Cancer patients who undergo stem cell transplantation face a long recovery, requiring medications with debilitating side effects and support around the clock. It’s a difficult experience, with studies showing that more than 70 percent of patients don’t adhere to drug regimens.  Statistician <a href="https://people.seas.harvard.edu/~samurphy/">Susan Murphy</a> spends her days trying to help people suffering from such challenging maladies. The Mallinckrodt Professor of Statistics and Computer Science and associate faculty at the <a href="https://kempnerinstitute.harvard.edu/">Kempner Institute</a> and her team address healthcare needs not through medicine, but by mobile apps. Murphy’s lab specializes in creating sophisticated computational instructions known as reinforcement learning algorithms, which form the technical backbone of next-generation programs to help people stick to a medication protocol, for instance, or regular tooth brushing, or reducing cannabis use.And if this sounds like one of those ubiquitous apps that tracks steps or counts calories, think again. “If you’ve ever downloaded a health app, those tend to be pretty dumb,” Murphy said. “For example, you’ll get a physical activity app, you’ll sprain your ankle, and it’ll continue to tell you to go for a walk.” “If you’ve ever downloaded a health app, those tend to be pretty dumb.” Susan MurphyUsing advancements in artificial intelligence and sensing technologies to move beyond one-size-fits-all interventions, the lab’s apps are capable of real-time personalization, meting out psychological rewards, and in some cases, leveraging social networks to help users stick to goals.This approach is called “<a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC5364076/">just-in-time adaptive intervention</a>” because it aims to provide support at just the right time by registering changing needs and contexts. Currently the Murphy lab is working with software engineers, cancer clinicians, and behavioral scientists to develop an app for stem-cell transplant patients and their primary caregivers, usually parents. Health management, especially for the sickest, typically requires involvement of others. For instance, up to 73 percent of family-care partners have primary responsibility for managing cancer-related medications. The researchers are in the early stages of developing the algorithm, to be deployed in a first-round clinical trial this year by collaborators at the University of Michigan and Northwestern University. The trial, called <a href="https://zipingxu.github.io/assets/AIME_Submission.pdf">ADAPTS HCT</a>, will focus on adolescent and young adult patients who’ve had stem-cell transplants in the 14 weeks post-surgery.The algorithm will inform sequential decisions, including when and whether to send motivational prompts to the patient, and whether to send messages and reminders to both patient and caregiver. The application includes a word-guessing game that fosters social support and collaboration between patient and caregiver. “We hypothesize that in improving the relationship between patients and their caregivers, patients can function and manage their medications better,” said Harvard postdoctoral fellow Ziping Xu, who is leading the ADAPTS HCT algorithm development.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1744676256060-120324_Susan_Murphy_Ziping_Xu_274.webp" style="width:100%px" class="fr-fic fr-dib" />Susan Murphy (left) and Ziping Xu. Photo by Grace DuValThe app will employ reinforcement machine learning, in which the software will “learn” from previous interactions. For example, rather than simply sending preset reminders about medications, the algorithm will tailor timing and content according to when they have been most useful to patients. That way there is less chance the notifications will be deemed irrelevant or ill-timed and eventually habitually ignored.“We use the algorithm to learn what is the best way to interact with each patient,” Xu said. “We use the algorithm to learn what is the best way to interact with each patient.”Ziping XuThe Murphy lab is deploying its algorithmic expertise across other domains. With their University of Michigan collaborators, they’ve recently pilot-tested a program called <a href="https://www.uofmhealth.org/MiWaves">MiWaves</a> aimed at young adults who are abusing cannabis. Like the ADAPTS HCT app, MiWaves continually learns and adapts from interactions with each patient to improve its decision rules, with the goal of helping them reduce their daily intake.The lab is also several years into a project called <a href="https://annatrella.github.io/research/w2024protocol/">Oralytics</a>, which recently wrapped up a 10-week randomized trial to help refine the delivery of push notifications to help patients adhere to a tooth-brushing protocol: two sessions of two-minute duration daily, covering all four mouth quadrants. The first Oralytics clinical trial included some 70 participants who all received the mobile app with a wireless-enabled toothbrush that sent data to the team’s collaborators at Proctor and Gamble. Graduate student Anna Li Trella, who led the Oralytics project through the first trial, said the recently collected data will help the team develop methods to better handle messy problems like missing data and software errors. “There are many constraints to running an algorithm in real life,” Trella said. “Now that we’ve conducted the first trial, we can make improvements to help the algorithm collect better data and learn better.”Murphy thinks of her lab as creating practical pocket coaches who can help people get where they want to go. “Very, very few people can afford a human coach. And in fact, some people may not want such intensive human interaction,” Murphy said. “That’s where the idea for these digital supports comes in.”

New apps for cancer patients, cannabis users, others make use of algorithms that continually customize support 

Like having a personal healthcare coach in your pocket

Healthcare is facing a growing workforce crisis. The World Health Organization estimates a global shortage of 10 million skilled healthcare workers by 2030 [1]. To maintain quality patient care, healthcare systems must leverage technology to augment clinicians, streamline workflows, and optimize resource allocation without replacing human expertise.Artificial Intelligence (AI) is emerging as a key enabler in this space. In 2024, AI-powered digital health companies received 42% of total global funding, up from just 7% less than a decade ago. However, AI adoption remains fragmented: while over a quarter of clinicians worldwide now use AI in their work, only 11% of healthcare organizations have multiple AI solutions in production [2]. This signals a clear demand for AI-driven efficiency but also reveals barriers to adoption, including integration challenges and resource-intensive AI models.Axelera AI addresses this challenge by empowering healthcare professionals with real-time, AI-driven insights at the edge. Its MetisTM AI Processing Unit (AIPU) enhances medical imaging diagnostics, treatment planning, and operational efficiency, helping clinicians reduce workloads, minimize diagnostic errors, and improve patient outcomes without replacing human decision-making. By delivering high-performance AI with low power consumption, Axelera AI is helping bridge the workforce gap and shape the next era of digital healthcare.<h2>Medical Imaging and Diagnostics</h2>Diagnostic errors pose a serious challenge in healthcare. Studies estimate that diagnostic errors contribute to 10% of patient deaths and up to 17% of adverse events in hospitals [3]. Many of these errors result from the sheer complexity of interpreting medical data, particularly in imaging, where radiologists analyze thousands of scans under time constraints.AI is helping to reduce these errors by identifying patterns in medical images that might not be immediately apparent to the human eye. Advanced machine learning algorithms help analyze large datasets, identify intricate patterns, and make remarkably accurate predictions about the course of diseases. AI-powered medical imaging tools use deep learning algorithms, convolutional neural networks, and computer vision techniques to analyze medical images with unprecedented precision.Axelera AI provides real-time AI inference processing for medical imaging, helping clinicians improve diagnostic accuracy and efficiency. Its Metis AIPU is designed to process large volumes of imaging data quickly while maintaining low power consumption, which is an important factor in resource-intensive hospital environments. By bringing this performance capability to the hospitals, doctors can also protect the patient’s identity and information by not having to share with a data center located offsite.<h3>Enhanced Imaging Analysis</h3>AI systems in medical imaging rely on deep learning models to process data from MRI, CT, X-ray, and ultrasound scans. These models can detect abnormalities such as tumors, fractures, or early-stage disease indicators. By enabling real-time image analysis, Axelera AI helps radiologists make informed decisions faster. In a radiology department, for example, Axelera AI can process MRI scans in real time, highlighting potential areas of concern, such as tumors or abnormal tissue growths. This allows radiologists to review flagged areas more efficiently and improve diagnostic confidence.<h3>Early Disease Detection</h3>Early detection is critical for conditions such as cancer, cardiovascular disease, and neurological disorders. AI models trained on vast medical datasets can identify subtle indicators of disease that may not yet be visible to the clinician’s eye. Axelera AI’s architecture supports these AI-driven detection models, allowing for high-speed analysis without requiring cloud processing. For instance, a healthcare provider can use Axelera AI to analyze mammograms, identifying early signs of breast cancer with higher precision. This supports timely intervention and better patient outcomes.<h2>Personalized Treatment and Precision Medicine</h2>AI is reshaping precision medicine by enabling data-driven, patient-specific treatment strategies. By analyzing vast datasets, AI algorithms identify patterns, predict individual treatment responses, and support clinicians in tailoring therapies for better patient outcomes. These models examine patient-specific data, including medical history, genetic information, lifestyle factors, and treatment outcomes of comparable patients.One of AI’s key strengths lies in integrating anonymized, diverse health data sources. AI-powered systems aggregate information from wearable technology, electronic health records (EHRs), imaging reports, and patient-reported outcomes. By leveraging computational modeling and real-time simulations, these systems can track treatment effectiveness dynamically, allowing healthcare providers to make adjustments as needed. This capability is particularly valuable in fields like oncology, where genetic variations and molecular signatures influence treatment success.Axelera AI brings high-speed, real-time AI processing to precision medicine applications. Its Metis AIPU is optimized for analyzing large-scale medical datasets on-device, reducing latency and eliminating reliance on cloud processing. By accelerating AI-driven genomic analysis and treatment simulations, Axelera AI enables clinicians to make faster, data-informed decisions for personalized care.<h3>AI-Driven Treatment Recommendations</h3>Precision medicine relies on identifying biomarkers and genetic markers that influence how a patient responds to specific treatments. AI models trained on large genomic datasets help physicians predict which therapies are most likely to succeed while minimizing adverse effects.Edge AI can enhance this process by rapidly analyzing patient-specific data, including genetic profiles and treatment history, to recommend optimal therapies. By running complex AI models directly at the edge, it delivers real-time insights without the processing delays associated with cloud-based systems.One use case is in oncology clinics. AI can analyze a patient’s genetic profile and medical history to suggest a personalized chemotherapy plan. It can also identify potential clinical trial options based on the patient’s specific biomarkers, expanding treatment opportunities beyond standard protocols.<h3>Predicting Drug Interactions and Side Effects</h3>AI can play a critical role in preventing adverse drug reactions by analyzing a patient’s medical history, existing medications, and genetic predispositions. AI-driven drug interaction models assess potential conflicts between prescribed drugs and provide recommendations for alternative medications that may be safer or more effective.Axelera AI’s high-throughput processing can enable real-time cross-analysis of patient medication data, minimizing the risk of harmful interactions. By leveraging AI inference at the edge, it can flag potential drug conflicts before prescriptions are finalized, ensuring proactive decision-making.A hospital pharmacy, for example, can integrate AI into its medication review system. When a physician prescribes a new medication, AI instantly scans the patient’s history, identifying potential interactions with existing treatments. If a risk is detected, it suggests alternative dosages or substitute medications, reducing the likelihood of adverse effects.<h2>Remote Patient Monitoring and Telemedicine</h2>The convergence of AI and telemedicine is driven by the need to overcome traditional healthcare challenges. With geographic barriers, clinician shortages, and rising healthcare costs, there is a growing need for solutions that reduce in-person visits while maintaining quality care. AI-powered remote patient monitoring (RPM) is an integral part of telemedicine, allowing healthcare providers to track patient health data in real time, detect early warning signs, and support virtual consultations more effectively.Axelera AI enhances RPM by providing real-time, AI-driven health monitoring at the edge. Unlike cloud-dependent solutions, its low-latency processing enables immediate detection of abnormalities, ensuring faster clinical responses. This is particularly beneficial for managing chronic conditions where continuous monitoring can prevent complications.Wearable devices and remote monitoring systems are now integral to chronic disease management, post-surgical recovery, and elderly care. AI-powered analysis of vital signs, heart rhythms, and other health metrics allows for timely alerts and early intervention.A patient with a chronic heart condition can wear a device that continuously monitors heart rate and rhythm. If an irregular heartbeat is detected, the system immediately alerts the patient’s cardiologist, prompting medical intervention before complications arise.Beyond monitoring, AI-driven telemedicine platforms enhance remote consultations by analyzing patient history, symptoms, and real-time data. This allows clinicians to make informed decisions without requiring an in-person visit.For instance, during a virtual consultation, Axelera AI would analyze a patient’s medical history and symptoms in real time, providing the physician with potential diagnoses and treatment recommendations. This enhances diagnostic accuracy while improving the efficiency of remote care.<h2>Operational Efficiency in Healthcare Facilities</h2>Hospitals and healthcare facilities operate under constant pressure to balance patient care, resource allocation, and administrative efficiency. Emergency room congestion, staff shortages, and unpredictable patient influxes create operational challenges that impact both costs and patient outcomes. AI-driven predictive analytics and workflow automation are helping healthcare providers manage resources more effectively, reduce administrative burdens, and optimize patient flow.Axelera AI enhances hospital efficiency by processing real-time operational data at the edge, enabling immediate decision-making without cloud dependency or the risky transmission of personalized data. Its high-speed AI inference capabilities allow hospitals to anticipate patient admission trends, optimize staffing, and automate resource allocation by training AI models on historical and real-time hospital data. This allows hospitals to proactively adjust staffing levels, manage bed availability, and allocate medical resources before shortages occur, which, in turn, enable them toA hospital using Axelera AI could, for example, predict an increase in emergency room admissions during flu season by analyzing trends in local infection rates, patient records, and historical data. This allows it to allocate additional staff and resources in advance, ensuring that patient care remains efficient and timely.Beyond predictive analytics, AI streamlines administrative and clinical workflows, reducing the burden of manual processes. Axelera AI enables automated scheduling, patient record management, and real-time prioritization of critical cases.For instance, in a busy clinic, Axelera AI can automate appointment scheduling, billing, and patient record updates. Instead of relying on manual coordination, the system dynamically adjusts schedules based on patient no-shows, appointment urgency, and physician availability. This allows staff to focus more on patient care rather than administrative tasks, improving overall efficiency.By integrating AI into hospital operations, healthcare providers can reduce inefficiencies, minimize administrative overhead, and improve resource utilization, leading to better patient experiences and lower operational costs.<h2>AI in Medical Research and Drug Discovery</h2>Developing new drugs is a slow, expensive, and resource-intensive process. Traditional methods rely on manual screening, lengthy clinical trials, and iterative testing, often taking years and requiring hundreds of millions of dollars in investment. Despite these efforts, many drug candidates fail in late-stage trials, delaying the arrival of effective treatments.AI is helping pharmaceutical companies overcome these bottlenecks by accelerating drug discovery, molecular analysis, and clinical research. Recent estimates suggest that AI-driven approaches can cut costs and reduce preclinical development timelines by as much as 25 to 50% [4]. With these efficiency gains, AI is expected to play a role in nearly one-third of new drug discoveries in 2025 [5]. This includes drug design, chemical synthesis, drug screening, and drug repurposing.<h3>Accelerating Drug Discovery</h3>AI-powered models can analyze biological and chemical datasets to identify promising drug candidates faster than conventional research methods. Machine learning algorithms help predict molecular interactions, toxicity risks, and drug efficacy, reducing the need for costly trial-and-error screening.Axelera AI enhances this process with real-time, AI-driven analysis at the edge. It accelerates large-scale computations, enabling researchers to simulate drug interactions efficiently without cloud-based delays.For example, a pharmaceutical company can identify new drug candidates for a rare disease by processing genetic and biochemical data using AI. AI-driven simulations can help the company cut development time and lower research costs, bringing new treatments to market faster.<h3>Genomic Data Analysis for Precision Medicine</h3>AI is also transforming genomic research, allowing scientists to identify genetic patterns linked to diseases and develop targeted therapies. By analyzing large genomic datasets, AI detects biomarkers and genetic variations, helping researchers predict how patients may respond to different treatments.Axelera AI processes genomic data at high speed, helping researchers analyze mutations, identify disease-linked genes, and refine precision medicine strategies. Its high-performance AI inference allows for real-time genomic analysis, reducing the computational workload of conventional sequencing methods.By integrating AI into drug discovery and genomics, AI helps researchers streamline drug development, reduce costs, and advance precision medicine.<h2>Enhancing Patient Experience and Care</h2>Patients today expect faster access to healthcare services, accurate information, and seamless communication with providers. However, long wait times, inefficient appointment scheduling, and fragmented patient records continue to create frustration. AI is helping healthcare providers bridge these gaps by enabling real-time patient interactions and improving data accuracy [6].<h3>AI-Powered Chatbots and Virtual Assistants</h3>By using Natural Language Processing (NLP) and machine learning, AI-driven chatbots and virtual assistants can interpret patient queries, automate scheduling, and provide instant access to health-related information. Beyond reducing response times and administrative workload, they enhance patient engagement by offering 24/7 accessibility. Axelera AI powers chatbots and virtual assistants with real-time AI inference at the edge. Unlike cloud-dependent solutions, it delivers low-latency processing, allowing AI-driven assistants to respond instantly, process multiple queries simultaneously, and provide context-aware answers.<h2>Conclusion</h2>Improving access to healthcare is a global concern. AI can significantly support this effort by supporting clinicians by making diagnostics more accurate, patient care more accessible, and hospital operations more efficient. Axelera AI plays a key role in this shift by enabling real-time AI processing for medical imaging, remote monitoring, and workflow automation. Its low-latency, edge-based AI capabilities help healthcare providers reduce diagnostic errors, improve care coordination, and accelerate medical research.As the industry continues to adopt AI-driven healthcare solutions, technologies like Axelera AI will be instrumental in scaling precision medicine, enhancing hospital efficiency, and supporting clinicians with real-time insights. By integrating AI into everyday healthcare workflows, providers can deliver faster, more personalized, and cost-effective care.<h2>References</h2>[1] A. Rivlin and T. Lumley, “Why is the world now facing a medical recruitment crisis?,” World Economic Forum. Jan. 9, 2023. [Online]. Available: <a href="https://www.weforum.org/stories/2023/01/medical-recruitment-crisis-davos23/">https://www.weforum.org/stories/2023/01/medical-recruitment-crisis-davos23/</a> [2] C. Stewart, “AI in healthcare,” Statista, Feb. 19, 2025. [Online]. Available: <a href="https://www.statista.com/topics/10011/ai-in-healthcare/#topicOverview">https://www.statista.com/topics/10011/ai-in-healthcare/#topicOverview</a>[3] RCSEd Communications Team, “The Potential of AI to Help Reduce Diagnostic Errors,” RCSEd. Sep. 3, 2023. [Online]. Available: <a href="https://www.rcsed.ac.uk/news-resources/rcsed-blog/2024/september/the-potential-of-ai-to-help-reduce-diagnostic-errors">https://www.rcsed.ac.uk/news-resources/rcsed-blog/2024/september/the-potential-of-ai-to-help-reduce-diagnostic-errors</a> [4] N. Sheynin, “How Artificial Intelligence is Transforming Drug Development,” AlphaSense. [Online]. Available: <a href="https://www.alpha-sense.com/blog/trends/expert-insights-artificial-intelligence-drug-development/">https://www.alpha-sense.com/blog/trends/expert-insights-artificial-intelligence-drug-development/</a> [5] P. Hudson, “How 2025 can be a pivotal year of progress for Biopharma,” World Economic Forum. Jan. 16, 2025. [Online]. Available: <a href="https://www.weforum.org/stories/2025/01/2025-can-be-a-pivotal-year-of-progress-for-pharma/">https://www.weforum.org/stories/2025/01/2025-can-be-a-pivotal-year-of-progress-for-pharma/</a> [6] S. Chowdhary, “Reimagining healthcare industry service operations in the age of AI,” McKinsey &amp; Company. Sep. 19, 2024. [Online]. Available: <a href="https://www.mckinsey.com/industries/healthcare/our-insights/reimagining-healthcare-industry-service-operations-in-the-age-of-ai">https://www.mckinsey.com/industries/healthcare/our-insights/reimagining-healthcare-industry-service-operations-in-the-age-of-ai</a>

Healthcare is facing a critical workforce shortage, with AI emerging as a key tool to support clinicians by improving diagnostics, treatment planning, and efficiency without replacing human expertise. 

Revolutionizing Digital Healthcare with Axelera AI: Enhancing Patient Care and Operational Efficiency

Chronic organ failure is a major cause of death – around 500,000 people are diagnosed with heart failure every year in the U.S. And yet, only a few thousand heart transplants are performed annually. Most people just aren't good candidates because of their age or condition and the scarcity of donor organs.At Stanford, I’ve spent the last five years pursuing a science fiction dream: What if we could 3D print new hearts and other organs on demand, using cells from a patient’s own body? How do we take a few cells in a petri dish and turn them into a kilogram of tissue that you can watch with your own eyes beating?There’s a lot of very hard challenges associated with scaling-up in biology. How do you make billions of cells affordably, reliably, and reproducibly? How do you print them quickly, mature them, let them thrive? We have to vascularize and feed the cells, put blood vessels into the tissues, so we can provide them the oxygen and nutrients they need to survive.In three and a half years, my lab plans to have a 3D-printed human organ inside a pig model. We hope that large animal study becomes a springboard to then lay out the milestones needed for a first in-human trial. That will likely be more than a decade away. There’s a lot of work to be done, but there are real forward steps that we're taking.The dream outcome is that we get a bioprinted heart in a human. With xenotransplants from pigs, you still need to be on a very high dose of immunosuppressants, and even then, it usually gets rejected after a month or two. Our goal would be to make your own heart, made of your own cells, on demand, with no need for immunosuppression. It really is the holy grail of what I think curative medicine ought to move towards: You’ve got a failing organ? Here’s a new one.I present this work to venture capitalists sometimes, and I talk to business people, and the response I get is that we’re being too ambitious. And to me, that means we’re doing academia right.I present this work to venture capitalists sometimes, and I talk to business people, and the response I get is that we’re being too ambitious. And to me, that means we’re doing academia right.”The private sector will fund what it’s willing to, which is five or 10 years in the future. When you’re really trying to do something different, train up a new workforce with a new set of skills, create new paradigms, that’s the role of academia. That’s our job. And it’s a privilege and an extremely exciting place to work. I just love my job, and I’m very grateful for the opportunity to take a problem that is impossible and make it not quite impossible.

Stanford bioengineer Mark Skylar-Scott on his “science fiction” quest to 3D print human hearts and other organs on demand, using cells from a patient’s own body.

'It really is the holy grail of curative medicine'

Polymer-coated nanoparticles loaded with therapeutic drugs show significant promise for cancer treatment, including ovarian cancer. These particles can be targeted directly to tumors, where they release their payload while avoiding many of the side effects of traditional chemotherapy.Over the past decade, MIT Institute Professor Paula Hammond and her students have created a variety of these particles using a technique known as layer-by-layer assembly. They’ve shown that the particles can effectively combat cancer in mouse studies.To help move these nanoparticles closer to human use, the researchers have now come up with a manufacturing technique that allows them to generate larger quantities of the particles, in a fraction of the time.“There’s a lot of promise with the nanoparticle systems we’ve been developing, and we’ve been really excited more recently with the successes that we’ve been seeing in animal models for our treatments for ovarian cancer in particular,” says Hammond, who is also MIT’s vice provost for faculty and a member of the Koch Institute for Integrative Cancer Research. “Ultimately, we need to be able to bring this to a scale where a company is able to manufacture these on a large level.”Hammond and Darrell Irvine, a professor of immunology and microbiology at the Scripps Research Institute, are the senior authors of the new study, which <a href="https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202503965" target="_blank">appears today in Advanced Functional Materials</a>. Ivan Pires PhD ’24, now a postdoc at Brigham and Women’s Hospital and a visiting scientist at the Koch Institute, and Ezra Gordon ’24 are the lead authors of paper. Heikyung Suh, an MIT research technician, is also an author.<img src="https://s3-us-west-1.amazonaws.com/wevolver-project-images/froala%2F1743722666957-MIT-Particle-Manufacturing-02-press.jpg" style="width:100%px" class="fr-fic fr-dib" />This microfluidic device can be used to assemble the nanoparticles rapidly and in large quantities. Credit: Gretchen Ertl<h2>A streamlined process</h2>More than a decade ago, Hammond’s lab developed a novel technique for building nanoparticles with highly controlled architectures. This approach allows layers with different properties to be laid down on the surface of a nanoparticle by alternately exposing the surface to positively and negatively charged polymers.Each layer can be embedded with drug molecules or other therapeutics. The layers can also carry targeting molecules that help the particles find and enter cancer cells.Using the strategy that Hammond’s lab originally developed, one layer is applied at a time, and after each application, the particles go through a centrifugation step to remove any excess polymer. This is time-intensive and would be difficult to scale up to large-scale production, the researchers say.More recently, a graduate student in Hammond’s lab developed <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC4847955/" target="_blank">an alternative approach</a> to purifying the particles, known as tangential flow filtration. However, while this streamlined the process, it still was limited by its manufacturing complexity and maximum scale of production.“Although the use of tangential flow filtration is helpful, it’s still a very small-batch process, and a clinical investigation requires that we would have many doses available for a significant number of patients,” Hammond says.To create a larger-scale manufacturing method, the researchers used a microfluidic mixing device that allows them to sequentially add new polymer layers as the particles flow through a microchannel within the device. For each layer, the researchers can calculate exactly how much polymer is needed, which eliminates the need to purify the particles after each addition.“That is really important because separations are the most costly and time-consuming steps in these kinds of systems,” Hammond says.This strategy eliminates the need for manual polymer mixing, streamlines production, and integrates good manufacturing practice (GMP)-compliant processes. The FDA’s GMP requirements ensure that products meet safety standards and can be manufactured in a consistent fashion, which would be highly challenging and costly using the previous step-wise batch process. The microfluidic device that the researchers used in this study is already used for GMP manufacturing of other types of nanoparticles, including mRNA vaccines.“With the new approach, there’s much less chance of any sort of operator mistake or mishaps,” Pires says. “This is a process that can be readily implemented in GMP, and that’s really the key step here. We can create an innovation within the layer-by-layer nanoparticles and quickly produce it in a manner that we could go into clinical trials with.”<h2>Scaled-up production</h2>Using this approach, the researchers can generate 15 milligrams of nanoparticles (enough for about 50 doses) in just a few minutes, while the original technique would take close to an hour to create the same amount. This could enable the production of more than enough particles for clinical trials and patient use, the researchers say.“To scale up with this system, you just keep running the chip, and it is much easier to produce more of your material,” Pires says.To demonstrate their new production technique, the researchers created nanoparticles coated with a cytokine called interleukin-12 (IL-12). Hammond’s lab has <a href="https://aiche.onlinelibrary.wiley.com/doi/10.1002/btm2.10453" target="_blank">previously shown</a> that IL-12 delivered by layer-by-layer nanoparticles can activate key immune cells and slow ovarian tumor growth in mice.In this study, the researchers found that IL-12-loaded particles manufactured using the new technique showed similar performance as the original layer-by-layer nanoparticles. And, not only do these nanoparticles bind to cancer tissue, but they show a unique ability to not enter the cancer cells. This allows the nanoparticles to serve as markers on the cancer cells that activate the immune system locally in the tumor. In mouse models of ovarian cancer, this treatment can lead to both tumor growth delay and even cures.The researchers have filed for a patent on the technology and are now working with MIT’s Deshpande Center for Technological Innovation in hopes of potentially forming a company to commercialize the technology. While they are initially focusing on cancers of the abdominal cavity, such as ovarian cancer, the work could also be applied to other types of cancer, including glioblastoma, the researchers say.The research was funded by the U.S. National Institutes of Health, the Marble Center for Nanomedicine, the Deshpande Center for Technological Innovation, and the Koch Institute Support (core) Grant from the National Cancer Institute.

MIT researchers Paula Hammond, Ivan Pires, and Ezra Gordon have developed a way to rapidly manufacture specialized nanoparticles that can be used for targeted delivery of cancer drugs and other therapeutics. Credit: Gretchen Ertl

Scaling up nanoparticle production could help scientists test new cancer treatments.

Engineers develop a way to mass manufacture nanoparticles that deliver cancer drugs directly to tumors

Nearly 85% of the 1.7 million adolescents with HIV live in sub-Saharan Africa, along with half of the nearly 40 million people in the world living with HIV. Although the government in Uganda provides antiretroviral treatment (ART) for free, adherence to the regimen by adolescents ages 10-16 is low, increasing the potential for the virus to further spread.<a href="https://sites.wustl.edu/rrtraining/people/claire-najjuuko/">Claire Najjuuko</a>, a doctoral student at Washington University in St. Louis, saw this firsthand while working as a data manager at the International Center for Child Health and Development (ICHAD) in Uganda, founded by <a href="https://brownschool.washu.edu/faculty-and-research/fred-ssewamala/">Fred M. Ssewamala</a>, the William E. Gordon Distinguished Professor in the Brown School at WashU. Now earning a doctorate in WashU’s Division of Computational &amp; Data Sciences, Najjuuko, who is co-advised by Ssewamala and <a href="https://engineering.washu.edu/faculty/Chenyang-Lu.html">Chenyang Lu</a>, the Fullgraf Professor in the Department of Computer Science &amp; Engineering in the McKelvey School of Engineering, wanted to use artificial intelligence and data science to help improve adolescent compliance with the treatment in low-resource areas. Results of the research were published online Feb. 25, 2025, in AIDS.“I have great interest in machine learning and want to apply it to problems that speak directly to me,” she said. “The collaborations between the AI for Health Institute directed by Professor Lu and the International Center for Child Health and Development directed by Professor Fred are particularly enabling this kind of innovative work.”With support from Lu and Ssewamala, Najjuuko set out to develop a machine learning model to predict which adolescents with HIV would be less likely to adhere to antiretroviral therapy. With such knowledge, health care practitioners could implement interventions for those identified as less likely to adhere to the treatment plan.“The current way the practice is, adolescents go to the clinic every month or two months for medication refills, and a health care practitioner checks how many pills the patient has left compared with what is expected, as well as asking the adolescent questions regarding missed doses to establish if the patient is adhering to the therapy,” Najjuuko said. “This project to predict future nonadherence of adolescents can have real impact if implemented in the right way.”To train the model, Najjuuko used data from a six-year cluster-randomized controlled trial from 39 clinics in southern Uganda, a region most heavily impacted by HIV. The Suubi+Adherence dataset included adolescents between age 10-16 medically diagnosed with HIV, aware of their status, enrolled in ART at one of the clinics and living within a family. Ultimately, the models analyzed data from 647 patients who had complete data on the outcome at 48 months.Najjuuko developed a machine learning model to predict nonadherence to antiretroviral therapy by incorporating socio-behavioral and economic factors alongside a patient’s adherence history. The model accurately identifies 80% of adolescents at risk of nonadherence while lowering the false alarm rate to 52% — 14 percentage points lower than a model based solely on adherence history. By reducing false alarms, this model helps health care providers focus interventions on those who need them most, improving patient outcomes while reducing unnecessary follow-ups and provider fatigue.Among 50 variables, which included social, interpersonal, family, educational, structural and economic factors, the model found 12 that were most predictive of an individual having poor adherence to ART. Economic factors were highly associated with future nonadherence. Other predictive characteristics were poor adherence history; child poverty; biological relationship to primary caregiver; self-concept; confidence in saving money; discussing sensitive topics with caregivers; household size; and school enrollment.“Adolescents are the most nonadherent group across the globe,” Ssewamala said. “They are moving into independence and don’t want to be told what to do. As they move into the dating period, there is a lot of stigma, and they don’t want to be associated with HIV.”One factor the team found was associated with adolescents with HIV adhering to the ART therapy was having a savings account.“The theory is when people own resources, especially when they have a nest egg, they think and behave differently,” Ssewamala said. “The future holds promise, so they will take care of themselves so they can live longer. When people are hopeless, they have nothing to lose.”Adhering to the treatment is difficult, Ssewamala said, because the medication must be taken with food or causes nausea. If a person with HIV doesn’t have access to food or transportation to get the medication, they are less likely to adhere to the treatment.Lu said this model could be adapted for deployment in the field to support personalized intervention strategies based on the identified risk factors, highlighting the importance of the collaboration.“This is an excellent example of interdisciplinary research at WashU, combining AI and global health,” Lu said. “By leveraging the data that Fred’s team gathered from the field and their insights on complex health issues, we apply AI expertise to analyze these data and build tools to enhance health outcomes.” <hr />Najjuuko C, Brathwaite R, Xu Z, Kizito S, Lu C, Ssewamala FM. Using machine learning to predict poor adherence to antiretroviral therapy among adolescents living with HIV in low-resource settings. AIDS. Online Feb. 25, 2025, DOI: <a href="https://journals.lww.com/aidsonline/abstract/9900/using_machine_learning_to_predict_poor_adherence.655.aspx">10.1097/QAD.0000000000004163</a>.Funding for this research was provided by the Eunice Kennedy Shriver National Institutes of Child Health and Human Development (NICHD) (1RO1HD074949-01).

Graduate student Claire Najjuuko set out to develop a machine learning model to predict which adolescents with HIV would be less likely to adhere to antiretroviral therapy. With such knowledge, health care practitioners could implement interventions for those identified as less likely to adhere to the treatment plan. (Credit: iStock photo)

Claire Najjuuko, Chenyang Lu, Fred Ssewamala use AI to target interventions to improve compliance

Machine learning could help predict adherence to HIV treatment in adolescents

In the modern era of sport, results are often decided by ultra-fine margins like milliseconds or millimeters. Yet even precise metrics such as speed and distance are still quite rudimentary in nature. The real wins for athletes, coaches, and sporting teams are gained through a much deeper understanding of sports performance and potential improvement. To achieve these goals and help obtain a competitive advantage, the latest performance tracking solutions lean on the power of wireless connectivity and edge processing. <h2>Monitoring unique performance metrics in sport </h2>Combining dozens of datapoints makes it possible to build a unique profile of an individual’s abilities. Examples in each sport might include the athlete’s body movement during key actions, or the position, launch angle, spin rate, and acceleration rate of the bat or ball at various game stages. For individual sports like athletics and cycling, key metrics might focus on biomechanical efficiency and technical precision. For team sports including the various football codes, common metrics include sprints, accelerations, decelerations, and workload intensity.  <h2>Bluetooth LE turns raw data into powerful insights </h2>Sports tracking solutions based on wireless tech make these and other powerful metrics readily available, in turn helping sports industry stakeholders achieve more nuanced insights. Tools like inertial sensors, anchor devices, and location services form systems that measure sports parameters in near real time. The high volume of raw data generated can be processed and analyzed using algorithms running on the sensor or wearable’s integrated wireless device before the information is forwarded—using low latency wireless connectivity such as <a href="https://www.nordicsemi.com/Products/Wireless/Bluetooth-Low-Energy">Bluetooth LE</a>—to an accompanying app on the user’s smartphone or a web-based dashboard on their laptop. <h2>Ball and athlete tracking technology in action </h2>Ball tracking technology, for example, delivers data on the trajectory, spin, and bounce of the ball, helping to measure the quality and accuracy of passes and shots, or evaluate defensive and offensive positioning. Athlete tracking technology, meanwhile, provides insights on the precise position, movement, and activity of the players themselves to help measure their performance, fitness, workload, and tactical decision-making. Moreover, wearable tech attached to the wrist, chest, or other parts of the body enables a much deeper dive into data, potentially revealing key information about an athlete’s heart rate variability (HRV), VO2 max, muscle fatigue, lactic acid, power output, and other important physical parameters. Such insights can provide a better understanding of athletic training, recovery, and injury prevention. <h2>A Nordic Bluetooth LE soccer tracking solution </h2>One prominent example of wireless sports technology in action is the Nordic-powered Apex Series, used by a host of major international sporting organizations including English Premier League football clubs.  Embedded in a custom lightweight vest worn by the athlete, STATSports’ <a href="https://www.nordicsemi.com/Nordic-news/2018/11/STATSports-APEX-device-employs-nRF52840-to-wirelessly-sync-comprehensive-sports-specific-metrics">Apex</a> employs Bluetooth LE connectivity to wirelessly sync sports-specific metrics to smartphones and tablets for coaches to make informed decisions during a match or training session. The device integrates a range of sensors including a high impact accelerometer, gyroscope, magnetometer, and augmented GNSS. These sensors provide comprehensive data including heart rate, total distance covered, current and maximum speed, number of sprints, accelerations, impacts, dynamic stress load, and step balance.  <h2>Edge AI for predictive modeling </h2>But the ultimate game-changer in the sports sector comes in the form of predictive modeling and data-driven analytics performed at the edge. When relevant data is not only crunched but analyzed and interpreted using the power of ‘<a href="https://www.nordicsemi.com/Products/Technologies/Edge-AI">Edge AI</a>’ algorithms, sports analysts can take advantage of even more precise and personalized insights. Adding AI and Machine Learning (ML) to the mix allows, for example, a sports wearable with the same sensors as before to now continuously monitor several performance parameters simultaneously, uncovering meaningful patterns that were previously hidden. By analyzing both historical data and real-time sensor information, statistical models can pinpoint factors that have a major impact on the way athletes prepare and perform, not to mention identify opposition strategies and gauge overall team dynamics. <h2>nRF54 Series SoCs enable the future of sports analytics devices </h2>Behind every breakthrough in next-generation sports tracking is hardware built to handle extreme demands. The fourth-generation wireless SoCs from Nordic Semiconductor, the nRF54 Series, deliver exactly that – combining unmatched processing power, ultra-low power consumption, and advanced on-chip AI capabilities.  To make sense of all the available sensor data in context and quickly enough to enable rapid and effective decision-making, powerful integrated wireless SoCs such as the nRF54L Series provide the serious processing heft to perform the ML and sensor fusion that forms part of Edge AI. By running Edge AI using a Nordic Semiconductor nRF54 Series SoC, for example, the local processing will use less power than sending data over the air, allowing a sports analytics device to have longer operation and smaller battery size.  The result is sophisticated sports analytics solutions that are changing the game.

In the modern era of sport, results are often decided by ultra-fine margins like milliseconds or millimeters. Yet even precise metrics such as speed and distance are still quite rudimentary in nature.

How next-gen wearables and Edge AI improve sports performance analytics

In medical device manufacturing, precision is not just a requirement—it’s the foundation of safety and performance. From diagnostic equipment to monitoring systems and surgical instruments, achieving high accuracy is essential. Among the various manufacturing methods, CNC machining has become indispensable for producing complex, high-precision components that meet the industry’s stringent demands. Let’s explore how CNC machining accelerates medical device development by delivering exceptional precision, optimizing production efficiency, and ensuring regulatory compliance.  <h2>I.      Overcoming Key Challenges in Medical Device Manufacturing</h2>Developing a medical device is a complex process—balancing tight tolerances, stringent regulations, and cost constraints demands precision at every step. Do these challenges sound familiar?High Costs for Small Batches: Developing medical devices often involves low-volume runs during prototyping and early production stages. Conventional manufacturing approaches, with their high setup costs and tooling requirements, struggle to offer cost-efficient solutions for these smaller quantities.Complexity in Precision Machining: The intricate designs and tight tolerances of medical components—such as thin-walled parts, internal cavities, and fine threads—demand exceptional precision. Achieving this consistency with traditional methods can lead to delays, rework, and increased expenses.Balancing Compliance with Efficiency: Adhering to regulatory standards like ISO 13485 is essential but maintaining compliance while achieving fast lead times and controlling costs remains a challenge.<h2>II. Why CNC Machining is a Game-Changer for Medical Device Development</h2>CNC machining transforms the way manufacturers produce medical device components, offering unparalleled precision, flexibility, and cost-efficiency throughout the product lifecycle. These advantages are key to delivering high-quality medical products while meeting strict regulatory standards.<h3>1.      Multi-Dimensional Precision Control</h3>CNC Machining provides exceptional accuracy and reliability through five key aspects:<ul><li>Dimensional Precision: CNC Machining can achieve remarkable precision. At RPWORLD, the machining precision can meet DIN ISO 2768 M (F), ensuring the exact specification for medical devices.</li><li>Surface Finish: Surface roughness levels as low as Ra 0.4μm can be achieved, though this depends on various factors such as part geometry, material selection, and tooling strategies.</li><li>Repeatability and Consistency: CNC machining ensures that every part in a production run adheres to exact specifications, reducing variability and guaranteeing uniform quality.</li><li>Complex Geometries: CNC machining supports intricate designs, including narrow grooves, thin walls, and internal cavities. Additionally, employing design techniques—<a href="https://www.rpworld.com/en/resources/blog/fast-cost-efficient-cnc-machining.html">check design guidelines here</a>—can further enhance manufacturability and maintain precision.</li><li>Stability Over Time: CNC machines are engineered for stable performance even during long production runs. This ensures that precision and quality do not degrade, making it possible to maintain strict and stable tolerances.</li></ul><h3>2.   Production Flexibility</h3>CNC machining supports quick, mold-free production, making it ideal for:• Prototyping: Enabling faster design validation and iterations with high-quality prototypes.• Small-Volume Production: Cost-effective for low-volume parts, allowing testing and refining without large-scale commitments.<h2>III. Meeting Stringent Quality &amp; Compliance Standards in CNC Machining</h2><h3>1.   Medical Device Classification</h3>Medical devices are classified into three categories based on their risk level and intended use. In the United States, the Food and Drug Administration (FDA) classifies medical devices into Class I, Class II, and Class III. In the European Union, medical devices follow a similar risk-based classification under the Medical Device Regulation (MDR):<ul><li>Class I: Low-risk devices such as diagnostic tools, bandages, and surgical instruments. These devices generally require minimal regulatory control.</li><li>Class II: Moderate-risk devices including patient monitors, infusion pumps, and imaging equipment. These devices require additional regulatory oversight, including performance standards and post-market surveillance.</li><li>Class III: High-risk devices, typically implantable or life-supporting, which require pre-market approval.</li></ul><h3>2.   The Role of ISO 9001: 13485</h3>ISO 13485 is the gold standard for quality management in medical device manufacturing. It ensures consistent quality, regulatory compliance, and patient safety by maintaining traceability, robust inspection protocols, and adherence to global standards. RPWORLD’s ISO 9001: 13485 certification demonstrates a commitment to maintaining the highest standards in CNC machining processes.<h2>IV. 5 Key Advantages of RPWORLD’s CNC Machining for Medical Device Manufacturing</h2>Extensive Industry Experience and ISO 13485 CertificationOver 20 years of expertise and ISO 13485 certification ensure compliance and reliability in CNC machining solutions for medical devices.Fast and Flexible TurnaroundWith lead times as fast as 3 days, RPWORLD helps meet tight project deadlines and adapt quickly to design changes.Precision and ComplianceDelivering high-precision components that meet DIN ISO 2768M (F) standards, ensuring the accuracy required for medical devices.Advanced Post-ProcessingOffering anodizing, plating, and laser engraving services tailored to medical needs, enhancing durability and functionality.Versatile Material OptionsProviding over 100 material choices, including medical-grade options such as PEEK and stainless steel, for diverse medical applications.

Let’s explore how CNC machining accelerates medical device development by delivering exceptional precision, optimizing production efficiency, and ensuring regulatory compliance.

How CNC Machining Accelerates Medical Device Manufacturing with Precision and Speed

In this episode, we explore a groundbreaking system developed by researchers at the Technical University of Munich that aids stroke patients in regaining arm and hand mobility.

Podcast: Exoskeleton + Digital Twin Aid Stroke Recovery

Personalized medicine, also known as precision medicine, represents an approach to healthcare that tailors medical treatment to the individual characteristics of each patient.

The Fusion of Precision Drug Delivery and Personalized Medicine

The body regulates its metabolism precisely and continuously, with specialised cells in the pancreas constantly monitoring the amount of sugar in the blood, for example. When this blood sugar level increases after a meal, the body sets a signal cascade in motion in order to bring it back down.In people suffering from diabetes, this regulatory mechanism no longer works exactly as it should. Those affected therefore have too much sugar in their blood and need to measure their blood sugar level and inject themselves with insulin in order to regulate it. This is a relatively imprecise approach compared to the body’s own mechanism.<h2>Equipping cells with special functions</h2>Martin Fussenegger is Professor of Biotechnology and Bioengineering at the Department of Biosystems Science and Engineering of ETH Zurich in Basel. With the above situation in mind, he and his team have been working on cell therapies for some time. One day, the hope is that these therapies will allow metabolic diseases such as diabetes to be treated individually and precisely – or even cured.But how do these cell therapies work? First, the researchers modify human cells by incorporating a network of genes that give the cells special abilities. These cells are <a>implanted</a> under the skin, for example, and the network is activated by a specific external stimulus.<h2>A suitable switch is key</h2>To that end, the researchers have developed various types of switches over recent years. Some can be controlled electrically, others with light, and one even using music by the British rock band Queen (see <a href="https://ethz.ch/en/news-and-events/eth-news/news/2023/08/cells-with-an-ear-for-music-release-insulin.html">ETH News</a>).The researchers in Basel have now developed another variant, which they have presented in the journal <a href="https://doi.org/10.1038/s41551-025-01350-7">Nature Biomedical Engineering.</a>“For me, this solution is the best gene switch that my group and I have built so far,” says Fussenegger. The reason is that the switch can be triggered using the long-established active ingredient nitroglycerine, and that the means of application – sticking a patch to the skin – is very simple. Corresponding patches are already available to buy in various sizes in any pharmacy.Nitroglycerine quickly diffuses out of the patch and into the skin, where it encounters an implant that contains modified human kidney cells.<h2>Network activated by nitric oxide</h2>These cells specifically intercept the nitroglycerine and have a built-in enzyme that converts it into nitric oxide (NO), a natural signalling molecule. In the body, NO normally causes blood vessels to dilate, leading to increased blood flow. It is broken down within a few seconds and therefore only affects a very localised area.The implanted cells are modified so that NO triggers the production and release of the chemical messenger GLP-1, which in turn boosts insulin release by the beta cells of the pancreas and thus regulates blood sugar level. GLP-1 also triggers a feeling of satiety, thereby reducing food intake.The new switch is made exclusively of human constituents – that is, it contains no components from other species. “That’s a new and groundbreaking feature,” says Fussenegger. With components from other species, there is always a risk of false triggering, interference with the body’s own processes, or immune reactions. “Here, we’re able to rule that out.”<h2>A whole arsenal of switches</h2>In the last 20 years, the ETH professor has developed various different gene switches, some of which respond to physical triggers such as current, sound waves or light. Which type has the best chance of being implemented one day?“Physical triggers are interesting because we don’t need to use molecules that interfere with the body’s own processes,” the biotechnologist says. He explains that electrical signals are ideal for controlling switches and gene networks using portable electronics such as smartphones or smartwatches – and AI can then be incorporated, too. “I therefore think electrogenetic cell therapies have the best chances of implementation. In terms of chemical switches, I see the new solution as being in pole position,” says Fussenegger.However, the further development of these cell therapies based on gene switches is a complex and lengthy process. “Developing a cell therapy to market maturity not only takes decades but also requires lots of staff and sufficient resources,” says the researcher. “There’s no shortcut.”Until now, Fussenegger’s work has focused mainly on cell therapies for diabetes, which is one of the world’s most prevalent metabolic diseases, affecting one in ten people. “That’s the model disease we work with. Fundamentally, however, it’s also possible to develop cell therapies for other metabolic, autoimmune or even neurodegenerative diseases – in principle, for everything that requires dynamic regulation.” According to Fussenegger, many drugs are like a hammer that is used to strike at a problem blindly. “Cell therapies, on the other hand, solve the problem in a similar way to the body,” he says.<h2>Reference</h2>Mahameed M, Xue S, Danuser B, Charpin-El Hamri G, Xie M, Fussenegger M: Nitroglycerin-responsive gene switch for the on-demand production of therapeutic proteins, Nature Biomedical Engineering, 14 February 2025, doi: <a href="https://doi.org/10.1038/s41551-025-01350-7">external page10.1038/s41551-025-01350-7</a>

A skin patch with nitroglycerine is the switch that controls an implant underneath (symbolic image). (Image: Josef Kuster / ETH Zurich)

ETH researchers have developed a new gene switch that can be activated using a commercially available nitroglycerine patch applied to the skin. One day, researchers want to use switches of this kind to trigger cell therapies for various metabolic diseases.

A new switch for the cell therapies of the future

In this episode, we explore how a generative AI tool is marking a major milestone in biology and accelerating advancements in healthcare, genetics, and drug development.<iframe height="200px" width="100%" frameborder="no" scrolling="no" src="https://player.simplecast.com/8b738738-5a69-4f97-ac33-8d48a50ec427?dark=false"></iframe> <hr /><h3>Episode Notes</h3>In this episode, we explore how a generative AI tool is marking a major milestone in biology and accelerating advancements in healthcare, genetics, and drug development.Become a founding reader of our newsletter: <a href="http://read.thenextbyte.com/">http://read.thenextbyte.com/</a><hr /><h2 id="isPasted">Transcript</h2>What's up folks, welcome back to the Next Byte podcast. And in this one, we're talking about DNA sequencing. Specifically, we're talking about a team over at Stanford that has created a ChatGPT-like AI model that can predict what comes next if you give it any DNA sequence, including yours.  that's got you a little bit spooked or excited or just curious. Well, then let's get into it.What's up friends, this is The Next Byte Podcast where one gentleman and one scholar explore the secret sauce behind cool tech and make it easy to understand.Farbod: This is another one of those biology ones. So, you already know I'm gonna struggle, but that's okay. We learn when we struggle. I was…Daniel: You just need to test your limits, man!Farbod: And that's what I'm doing. I'm trying here, but it was so interesting that I knew we needed to talk about it. So, for the audience, please bear with me. This is not my area of expertise, but God am I trying. So, for this episode, we're going to Stanford. You know the West Coast is the best coast, but these folks at Stanford, they're doing some crazy things with DNA sequencing.  And to be completely honest, they've kind of done this before, but now they're doing it better. So, what am I talking about? They've developed a chatGPT-like model, right? It's structured like a large language model, but instead of processing the meaning of the word that you're typing or the paragraph that you're writing or whatever and then giving you an answer or a follow-up or completing your train of thought, it takes as an input. A sequence of nucleotides, a sequence of DNA, and it completes it for you or it makes sense of it and it tells you what genes might be associated with each other. Now, before we get into that, I think we got to take it a couple steps back and by we, I mean me because I know I need to take it, a couple steps back. So, they did an amazing job in this article, by the way, explaining things like you're a five-year-old, which again is wonderful for me, and they have a three-minute video that I highly recommend you check out. It's going to be attached to the show notes. But they explained the issue at hand like this. They said, imagine you're reading the book, and this book has three billion letters in it, right? And every one of those letters is a nucleotide. These nucleotides make up the DNA that everything we know of in this world is pretty much composed of. And it's actually kind of straightforward. We represent them as four letters, A, C, G, and T. And standalone, they don't really mean much, but the sequence of them in different patterns together make up specific genes. Now, as we know, if you've been watching the news, this gene causes this, and poor, broken nucleotide sequences can, for most people, be harmless. For some people, it can cause cancer. So, there's a lot of value in understanding what those sequences mean, how they're related to each other, et cetera, et. Now, as I mentioned, this team had already worked on a chatGPT-like model before called the EVO-1 that was trained using 300 billion nucleotides, mostly from bacteria and stuff like that, to help them proof of concept this idea of, if we had a model that could predict what comes next. Could we better understand what the next evolutionary step would be for this plant or these bacteria or this human? Or could we better understand how the sequence that is at the one millionth element is related to the one at the 100,000th element? And if so, what does that tell us about the chances that you might have cancer? Or could we edit this out? The possibilities and the downsides of what already exists, basically.Daniel: Well, and when we talked about the technological development of having sequenced the human genome several decades ago, that didn't necessarily crack the code for us on what everything in human DNA means. That was just us looking at the DNA and writing a list, writing that three billion letter long book of just A’s, C’s, G’s, and T’s. It's awesome that like there's only four letters, but essentially, we didn't understand how those letters comprise genes and genes, they liken it in the article to like words. So, like basically, the major development in sequencing the human genome was like figuring out the alphabet and the order of the letters in the alphabet. But we did not yet understand the vocabulary. We didn't understand the words. And beyond that, we don't understand the grammar about how you put several words together to create phrases. And that's how physical characteristics are expressed or through a combination of several different genes. So, this is essentially helping us bridge the gap. And like in my mind, truly cracked the code on the meaning of DNA. If we can start to predict protein form and function by looking at the order of the A's, C's, G's and T's, predict harmful slash beneficial genes, potentially create new genetic sequences for medicines or treatments like this. This is bridging the gap between the letters in the physical realm, whereas before it was awesome that we understood what word of the letters were in.Farbod: You know, just like EVO2 is making sense of these nucleotide sequences, I'm happy you're here to make sense of my train of thought about biology.Daniel: EVO2 wouldn't exist without EVO1, buddy.Farbod: You're right. And like, I guess that frame of thinking is even better when comparing it to something like ChatGPT or really any other large language model, because the key breakthrough there, I think we mentioned it five episodes ago. The key breakthrough there was the transformer architecture, which said in relation to this wider body of text, what does this letter mean and this letter mean, instead of trying to make sense of a sequence of words back-to-back, which then lost meaning once you reach the 50th word. So again, they've taken the same underlying engine here, but now applying it to what does this sequence of genes mean, these nucleotides mean. And I think, I mentioned that EVO1 had 300 billion nucleotides, which already sounds quite impressive, but it was mostly bacteria and whatnot. Well, they upped it for EVO2, and now they include animals, plants. Overall, there's nine trillion nucleotides that are represented in this model. The only thing that they left out by design was viruses. The team said they didn't want this tool to be something that bad actors could use to develop and refine existing viruses, which I personally thank them. I think no one wants another round of the coronavirus. So, kudos for thinking ahead on that.Daniel: And this is like, it's pretty monumental here. So, in addition to like excluding viruses, right? Basically, except for viruses, the training data now is 27 times bigger than before. It includes all known living species that we have a DNA sequence for, plus some that are extinct. So, it's like, it's pretty much everything that we've ever sequenced ever, except for viruses so that we don't have people going, creating designer viruses to go try and get people sick. But every living being that we have the DNA sequence for has been included in the training data for this next generation. That's a big deal to increase your training data size, your training data set by a factor of 27, right? It's not like they've trained on 10 % more sequences than before. This is a huge deal.Farbod: Absolutely. And not only have they increased the number of data points that they're training on, but the input is also, what does ChatGPT call them, the number of tokens that you can input for a query? Yeah.Daniel: Yeah.Farbod: So, they've upped it to a one million letter sequence of nucleotides. So, you can have one million letters of combination of ACGT and it can then do whatever you want with it, analyze it, predict the next set of sequences, cetera, et. And I was wondering why one million? Apparently in the world of biology, the one million mark is, it's critical because it's long enough that it takes into account historically significantly far enough genes to make sense of rather than just taking into account the recent developments or deficiencies.Daniel: That makes sense. I appreciate the significance around there. And this makes me feel more so.  ChatGPT is an awesome analogy to give people, especially because I feel like a lot of people have had exposure to ChatGPT. But I'm hoping within our audience, maybe there's folks who've had a greater exposure to things like a co-pilot in coding or things like cursor composer. This actually feels a lot like cursor composer to me where in the cursor editor, you can go in the composer bar and you can have your code base living there, which is like a sequence of DNA. And you can tell it, hey, I want to generate a change to do this to the code  and composer will go and make those changes based off of its understanding of the code. This feels a lot like that, right? So, some of the technical features that it has, similar actually again to cursor composer is like autocomplete. They make a big deal about it in cursor. They call them tabs, but basically like you're writing and it understands what you're trying to code and you can just hit tab and it completes it, right? Autocomplete for code. They have autocomplete for DNA here based off of patterns in DNA sequences. They also have prediction models that can understand real world function. So instead of saying like change all the A's to C's or you know, control F my ACGTs and change them to TCGAs, you can say, I actually just want to change how this physical characteristic appears in the real world. And then it'll go change the DNA to reflect that. And then the interesting part, which is really awesome in my mind, it tends to be somewhat controversial. But we've seen a lot of work with CRISPR gene editing, which basically allows us to like sequence DNA, fabricate it, and then inject it into a DNA sequence. And then in a living being and basically edit a living being's DNA. They're able to essentially predict what the outcome of CRISPR gene editing will look like in the real world to propose edits essentially propose future experiments and tests.Farbod: Yes. That's what I gonna bring up I think, you know, the audience might have mixed feelings about that but what stood out to me? They were like look within this model with an EVO2 we can whenever you're predicting a new gene, we can tell you if it already exists if it doesn't if you implement it like what kind of behavior it could have, and then obviously you can do lab testing, put it into a living cell and see how it pans out. One thing that I thought was interesting in regards to CRISPR and gene editing, they were like, what if we could use this tool to modify bacteria to consume microplastics and to clean up oil spills, right? So again, I know it's a mixed feeling type of topic, especially as it relates to human beings, but the benefits of this tool are actually pretty significant. And even on the human-Daniel: This doesn't necessarily dictate how the findings are applied, right? This is an engine for understanding DNA, right? That's what we should be excited about, is it's an engine for us to understand what's going on. Even if you decide not to go edit your genes, to get rid of a disease, you can understand what diseases you might be predisposed to based off of your DNA.Farbod: Yeah, and you know, most of the stuff that they talked about are all the positives that we can all agree on, like modifying the bacteria. They said this can be used to significantly cut down on the discovery phase of new drug development because you can find which proteins work best for addressing whatever symptom. We talked about these two episodes ago, I think, on the snake venom episode, which has been a hit, by the way. They did mention that EVO2 could help with human evolution because we could predict what genes could be beneficial to us, but that's a topic for another day. I think what we should focus on here is that this is an incredible platform that they have open source. It's available to all researchers that they want others to use in their workflow and build on top of so that we can make better drugs and help our environment.Daniel: Well, and I think two really important things to mention. And you mentioned one of them and I kind of want to double down on it. One of them is direct quote from the researchers. They said it's really, really good at drug discovery, which is awesome. So obviously, helping identify treatments for potential diseases is awesome. But one of the other things that they can mention is like they, this model can distinguish between harmless DNA mutations to disease, disease causing ones. So, where we've got a lot of concern around free radicals and/or mutations in your DNA, potentially cancer causing, as you mentioned before, is like, we're trying to limit exposure to anything that could cause any type of mutation in your DNA. And that's where we get a lot of conflicting advice from people like, oh, like don't put on sunscreen, but also don't get sunburned and don't eat this and don't eat that. And don't do that. Like it feels like you're, you're screwed either way. I feel like this will potentially help us get more clarity around which types of causes for DNA mutations are most likely to cause disease and then help us hone our focus on like, what's the 80-20? What's the Pareto principle? Like what's the 20% of effort that I can do to protect myself from 80% of bad outcomes? I feel like that's one of the most important things that you and I will experience in our day-to-day life is a little bit more clarity around a lot of the conflicting guidance that we get. I think the most specific one that I think of is like around sun exposure. Like they're like, you need sun exposure for vitamin D, but don't get so much sun exposure that it causes mutations in your skin can cause melanoma. So, if you're outside, make sure you put on sunscreen, but also sunscreen can like cross through the blood brain barrier and cause disruptions in your hormones. Try and give us some clarity around a lot of the conflicting advice like that, where it feels like you can't win either way. Like, hopefully this will help us understand more what changes to make in a day-to-day life that actually have an impact on the presence of disease in our DNA.Farbod: Dude. Yeah, I totally agree that was a good hot take that I completely missed. And the sunscreen one, I'm totally with you. I think about it every morning when I put on my sunscreen. With that said, I'm going to wrap things up. So, folks, if you've heard about DNA sequencing and what a miracle it was that we cracked it a couple decades ago, you might be shocked to hear that it didn't crack the code all the way. Look, we understand what the nucleotides are that make up our DNA. You know, the four different ones, the A, C, G and T. But these folks over at Stanford are now developing a model or have already developed a model that can make sense of the different genes that are part of this DNA sequence and how they interact with each other. But more importantly, almost like you're using ChatGPT, it can predict what comes next. Now, why is this valuable? Well, by understanding the relation between the different genes, they can understand what is harmful, what can be used to develop new tools, how we can modify an existing sequence to give us beneficial properties. For example, one they called out was modifying bacteria to eat microplastics or clean up oil spills. And the next bit, it could even help us understand what good way we can go for evolving the human genome so that we get all the best properties possible.Daniel: That's Pretty fireman.Farbod: Sweet. That's the pod.Daniel: All right. See you.<hr />As always, you can find these and other interesting &amp; impactful engineering articles on Wevolver.com.To learn more about this show, please visit our<a href="https://www.wevolver.com/profile/the.next.byte"> shows page</a>. By following the page, you will get automatic updates by email when a new show is published. Be sure to give us a follow and review on<a href="https://podcasts.apple.com/nl/podcast/the-next-byte/id1548255861?l=en" target="_blank" rel="nofollow"> Apple</a> podcasts,<a href="https://open.spotify.com/show/6lPPsRtegnivsRUEsQ09R7?si=IPMiah7xT_apJtPjqvXMlA" target="_blank" rel="nofollow"> Spotify</a>, and most of your favorite podcast platforms!--The Next Byte: We're two engineers on a mission to simplify complex science &amp; technology, making it easy to understand. In each episode of our show, we dive into world-changing tech (such as AI, robotics, 3D printing, IoT, &amp; much more), all while keeping it entertaining &amp; engaging along the way.

In this episode, we explore how a generative AI tool is marking a major milestone in biology and accelerating advancements in healthcare, genetics, and drug development.

Podcast: Stanford Predicts The Next Human Evolution

<html><head></head><body>Spinal cord injuries are life-altering, often leaving individuals with severe mobility impairments. While rehabilitation robotics—devices that guide movement during therapy—have improved training for those with spinal cord injuries, their effectiveness remains limited. Without active muscle engagement, robotic-assisted movement alone does not sufficiently retrain the nervous system.<iframe width="640" height="360" src="https://www.youtube.com/embed/KnyHjGcDnME?&amp;wmode=opaque&amp;enablejsapi=1&amp;origin=https://www.wevolver.com" frameborder="0" allowfullscreen="" class="fr-draggable" data-lf-form-tracking-inspected-kn9eq4roawkarlvp="true" data-lf-yt-playback-inspected-kn9eq4roawkarlvp="true" data-lf-vimeo-playback-inspected-kn9eq4roawkarlvp="true"></iframe> A team at .<a href="https://www.neurorestore.swiss/" target="_blank" rel="noopener">NeuroRestore</a>, led by Grégoire Courtine and Jocelyne Bloch, has now developed a system that seemlessly integrates an implanted spinal cord neuroprosthesis with rehabilitation robotics. The researchers’ device delivers well-timed electrical pulses to stimulate muscles in harmony with robotic movements, resulting in natural and coordinated muscle activity during therapy. The neuroprosthetics innovation leveraged the the robotic expertise of <a href="https://www.epfl.ch/labs/biorob/">Professor Auke Ijspeert’s lab at EPFL</a>. This advancement not only enhances immediate mobility but also fosters long-term recovery.“The seamless integration of spinal cord stimulation with rehabilitation or recreational robotics will accelerate the deployment of this therapy into the standard of care and the community of people with spinal cord injury,” says Courtine. This adaptability ensures that rehabilitation professionals can incorporate this technology into existing rehabilitation protocols worldwide. Combining therapies also presents significant challenges, as each requires precise synchronization. Spinal cord stimulation strategies must be modulated in both space and time to match the patient’s movement, and integrating them with widely used robotic rehabilitation systems requires a flexible and adaptable framework.The technology relies on a fully implanted spinal cord stimulator that delivers biomimetic electrical epidural stimulation (electrical epidural stimulation). Unlike traditional functional electrical stimulation, this method activates motor neurons more efficiently by mimicking natural nerve signals.The researchers integrated electrical epidural stimulation with various robotic rehabilitation devices—including treadmills, exoskeletons, and stationary bikes—ensuring that stimulation is precisely timed with each phase of movement. The system uses wireless sensors to detect limb motion and automatically adjust stimulation in real time, allowing for a seamless user experience.In a proof-of-concept study involving five individuals with spinal cord injuries, the combination of robotics and electrical epidural stimulation resulted in immediate and sustained muscle activation. Not only did participants regain the ability to engage muscles during robotic-assisted therapy, but some also improved their voluntary movements even after the stimulation was turned off.The researchers also worked closely with rehabilitation centers to test how well the stimulation system integrated with widely used robotic devices. “We visited multiple rehabilitation centers to test our stimulation technology with the robotic systems they routinely use, and it was incredibly rewarding to witness their enthusiasm,” say .NeuroRestore researcher Nicolas Hankov and BioRob researcher Miroslav Caban, the study’s first authors. “Seeing firsthand how seamlessly our approach integrates with existing rehabilitation protocols reinforces its potential to transform care for people with spinal cord injury by providing a technological framework that is easy to adopt and deploy across multiple rehabilitation environments.”The study also showed the potential of this approach beyond clinical settings, as participants used the system to walk with a rollator and cycle outdoors, validating its real-world impact.This innovative technology offers new hope for individuals with spinal cord injuries, presenting a more effective rehabilitation approach than robotics alone. By making rehabilitation more dynamic and engaging, it has the potential to significantly enhance recovery outcomes. Future clinical trials will be needed to establish long-term benefits, but the initial results suggest that integrating neuroprosthetics with rehabilitation robotics could redefine mobility restoration after paralysis.List of contributors<ul><li>EPFL Neuro X</li><li>Lausanne University Hospital (CHUV) and University of Lausanne (UNIL)</li><li>Defitech Center for Interventional Neurotherapies (.NeuroRestore)</li><li>EPFL Biorobotics Laboratory</li><li>ONWARD Medical</li><li>Bern University of Applied Sciences</li><li>VAMED Management and Service Switzerland AG</li><li>ETH Zurich Sensory-Motor Systems Lab</li><li>University of Zurich Spinal Cord Injury Center</li><li>Hocoma AG</li><li>Medtronic</li><li>Oxford University</li><li>GBY (Go-by-Yourself) SA</li><li>Université de Bordeaux Institut des Maladies Neurodégénératives</li><li>Zurich University of Applied Sciences (ZHAW)</li><li>MyoSwiss AG</li></ul><a href="https://www.dropbox.com/scl/fo/mrs3q6p0swmb2ojq4hz9r/APSxvOKXuBsk7FztEVNrhZA?rlkey=im02yangj4ghnadhi7g7eqvei&amp;e=1&amp;st=4p8m3tfd&amp;dl=0" target="_blank" rel="noopener">Press kit</a><hr>ReferencesNicolas Hankov, Miroslav Caban, Robin Demesmaeker, Margaux Roulet, Salif Komi, Michele Xiloyannis, Anne Gehrig, Camille Varescon, Martina Rebeka Spiess, Serena Maggioni, Chiara Basla, Gleb Koginov, Florian Haufe, Marina D’Ercole, Cathal Harte, Sergio D. Hernandez-Charpak, Aurelie Paley, Manon Tschopp, Natacha Herrmann, Nadine Intering, Edeny Baaklini, Francesco Acquati, Charlotte Jacquet, Anne Watrin, Jimmy Ravier, Frédéric Merlos, Grégoire Eberlé, Katrien Van den Keybus, Hendrik Lambert, Henri Lorach, Rik Buschman, Nicholas Buse, Timothy Denison, Dino De Bon, Jaime E. Duarte, Robert Riener, Auke Ijspeert, Fabien Wagner, Sebastian Tobler, Leonie Asboth, Joachim Von Zitzewitz, Jocelyne Bloch, Grégoire Courtine. Augmenting rehabilitation robotics with spinal cord neuromodulation: a proof of concept. Science Robotics 12 March 2025. DOI:&nbsp;<a href="https://doi.org/10.1126/scirobotics.adn5564" target="_blank" rel="noopener">10.1126/scirobotics.adn5564</a></body></html>

Common robotic devices to safely automate and augment gait rehabilitation. © .NeuroRestore/EPFL/CHUV 2025

Scientists at .Neurorestore (EPFL/CHUV/UNIL) have developed an approach that combines rehabilitation robotics with spinal cord stimulation to restore movement in people with spinal cord injuries. The technology enhances rehabilitation and enables activities like cycling and walking outdoors.

Robotics and spinal stimulation restore movement in paralysis

This article was discussed in our <a href="https://www.wevolver.com/profile/the.next.byte" rel="noopener noreferrer">Next Byte podcast</a>. <iframe height="200px" width="100%" frameborder="no" scrolling="no" src="https://player.simplecast.com/f1afc471-449b-4578-a9d3-c68713d77587?dark=false"></iframe> The full article will continue below.<hr />The researchers use functional electrical stimulation (FES) to stimulate specific muscles in the forearm. This is necessary, for example, for moving fingers, grasping objects or catching a ball. However, one-sided paralysis following a stroke usually affects not only the hand, but the entire side of the body. For that reason, a scaffold also supports the entire arm up to the shoulder.<h2>Modular system with computer game: independent training</h2>Twenty-four stroke patients have already used the complete system, consisting of an exoskeleton for the arm and shoulder in combination with FES as part of the ReHyb research project. Half of them were patients at the Schön Klinik Bad Aibling Harthausen, which is leading the study. The researchers also used a computer game that automatically adapts to the individual player’s capabilities. It trains them to grip and move their arms shortly after a stroke by reacting to colored balls flying toward them at varying speeds on a screen. The task is to catch the balls and match them with color-coded boxes.<h2>The secret of success: digital twin of muscle activity, muscle stimulation strength and exoskeleton</h2>At the center of TUM Professor Sandra Hirche’s setup is a digital twin that records the individual requirements of each patient and places them in a control loop. Among other things, the researchers have to determine how well each patient can move their arm and hand. In the event of a stroke, for example, paralysis can be caused by damage to the motor area in the brain responsible for movement. However, it is impossible to predict how severely the signals transmitted from the brain to the muscles in the forearm will be impaired after the stroke. “Individual muscle strands in the forearm can be stimulated to the right extent for hands and fingers to move,” says Prof. Hirche, who holds the Chair of Information-Oriented Control at TUM. In addition to information on muscle activity in the forearm, the researchers need to know how strongly the muscles should be stimulated in conjunction with the exoskeleton assistance. “We use algorithms to bring this individual information together in a control loop,” says the control engineering expert. Consequently, the digital twin is needed to provide individualized support for the arm and hand movements of affected persons.<h2>Schön Klinik: modular system as a home trainer</h2>Prof. Hirche uses the phrase “intention-controlled intelligent control” to refer to the fact that patients can use this technology to move as much as they want after a stroke. Carmen Krewer, team lead of the research group at the Schön Klinik cooperation partner in Bad Aibling, enthuses: “Such a modular system with electrical stimulation and exoskeleton has never existed. It also enables stroke sufferers to continue training at home without the support of others.”<iframe width="640" height="360" src="https://www.youtube.com/embed/OboVMCkM9Yk?&amp;wmode=opaque&amp;rel=0&amp;enablejsapi=1&amp;origin=https://www.wevolver.com" frameborder="0" id="366335023"> </iframe><hr /><h3 id="isPasted">Publications</h3>Framework for Learning a Hand Intent Recognition Model from sEMG for FES-Based control; N. Das, S. Endo, H. Kavianirad, S. Hirche; 2024; <a href="https://ieeexplore.ieee.org/document/10719910" target="_blank" rel="noreferrer">ieeexplore.ieee.org/document/10719910</a>

Wolfgang Korisanski is one of the 24 patients who have already tested the new exoskeleton and electrical stimulation system for a study. In the background: Hossein Kavianirad (left) and Neha Das, researchers from Prof Sandra Hirche's department

Researchers at the Technical University of Munich (TUM) have developed a system that helps patients learn to move their paralyzed arms and hands quickly after a stroke. This requires targeted stimulation of the muscles in the forearm and the support of an exoskeleton.

Regaining mobility quickly after a stroke

<html><head></head><body>Millions of Americans undergo surgery each year. After surgery, preventing complications like pneumonia, blood clots and infections can be the difference between a successful recovery and a prolonged, painful hospital stay – or worse. More than 10% of surgical patients experience such complications, which can lead to longer stays in the intensive care unit (ICU), higher mortality rates and increased health care costs. Early identification of at-risk patients is crucial, but predicting these risks accurately remains a challenge.New advancements in artificial intelligence (AI), particularly large language models (LLMs), now offer a promising solution. A recent study led by&nbsp;<a href="https://engineering.washu.edu/faculty/Chenyang-Lu.html">Chenyang Lu</a>, the Fullgraf Professor in computer science &amp; engineering in the McKelvey School of Engineering and director of the&nbsp;<a href="https://aihealth.wustl.edu/">AI for Health Institute</a> (AIHealth) at Washington University in St. Louis, explores the potential of LLMs to predict postoperative complications by analyzing preoperative assessments and clinical notes. The work, published online Feb. 11 in&nbsp;<a href="https://www.nature.com/articles/s41746-025-01489-2">npj Digital Medicine</a>, shows that specialized LLMs can significantly outperform traditional machine learning methods in forecasting postoperative risks.“Surgery carries significant risks and costs, yet clinical notes hold a wealth of valuable insights from the surgical team," Lu said. "Our large language model, tailored specifically for surgical notes, enables early and accurate prediction of postoperative complications. By identifying risks proactively, clinicians can intervene sooner, improving patient safety and outcomes.”Traditional risk prediction models have primarily relied on structured data, such as lab test results, patient demographics, and surgical details like procedure duration or the surgeon’s experience. While this information is undoubtedly valuable, it often lacks the nuance of a patient's unique clinical narrative, which is captured in the detailed text of clinical notes. These notes contain personalized accounts of the patient's medical history, current condition, and other factors that influence the likelihood of complications.Lu and co-first authors Charles Alba and Bing Xue, both graduate students working with Lu at the time the study was conducted, employed specialized LLMs trained on publicly available medical literature and electronic health records. They then fine-tuned the pretrained model on surgical notes to make better predictions about surgical outcomes. The resulting method – the first of its kind to process surgical notes and use them to make predictions about postoperative outcomes – can go beyond structured data to recognize patterns in the patient's condition that might otherwise be overlooked.&nbsp;Based on nearly 85,000 surgical notes and associated patient outcomes from an academic medical center in the Midwest collected between 2018 and 2021, the team reported that their model performed far better than traditional methods in predicting complications. For every 100 patients who experienced a postoperative complication, the team’s new model correctly predicted 39 more patients who had complications than traditional natural language processing models.Beyond the number of patients who could potentially have surgical complications caught early and mitigated, the study also showcases the power of foundation AI models, which are designed to multitask and can be applied to a wide range of problems.“Foundation models can be diversified, so they’re generally more useful than specialized models. In this case, where lots of complications are possible, the model needs to be versatile enough to predict many different outcomes,” said Alba, who is also a graduate student in WashU’s&nbsp;<a href="https://datasciences.wustl.edu/">Division of Computational &amp; Data Sciences</a>. “We fine-tuned our model for multiple tasks at same time and found that it predicts complications more accurately than models trained specifically to detect individual complications. This makes sense because complications are often correlated, so a unified foundational model benefits from shared knowledge about different outcomes and doesn’t have to be painstakingly tuned for each one.”“This versatile model has the potential to be deployed across various clinical settings to predict a wide range of complications,” said&nbsp;<a href="https://anesthesiology.wustl.edu/people/joanna-abraham-phd/">Joanna Abraham</a>, associate professor of anesthesiology at WashU Medicine and a member of the Institute for Informatics (I2) at WashU Medicine. “By identifying risks early, it could become an invaluable tool for clinicians, enabling them to take proactive measures and tailor interventions to improve patient outcomes.”&nbsp;<hr>Alba C, Xue B, Abraham J, Kannampallil T, Lu C. The foundational capabilities of large language models in predicting postoperative risks using clinical notes.&nbsp;njp Digital Medicine, published online Feb. 11, 2025. DOI:&nbsp;<a href="https://www.nature.com/articles/s41746-025-01489-2">https://www.nature.com/articles/s41746-025-01489-2</a>&nbsp;</body></html>

Study led by Chenyang Lu explores the potential of large language models to predict postoperative complications by analyzing preoperative assessments and clinical notes. (Photo: iStock)

AI for Health Institute team unveils a versatile large language model to enhance perioperative care

Foundation AI model predicts postoperative risks from clinical notes

The future of healthcare is shifting beyond hospital walls, bringing intelligent, power-efficient health monitoring solutions directly into patients’ daily lives. But making remote patient monitoring (RPM) truly effective requires overcoming major barriers: battery life, device size, and data privacy. In this webinar, host Samir Jaber is joined by Carlos Morales, VP of AI at Ambiq, in a discussion on how Ambiq’s proprietary Subthreshold Power Optimized Technology (SPOT®) and ultra-low-power semiconductor solutions are pushing the envelope on body-worn AI and RPM solutions. Learn how Ambiq is tackling industry-wide challenges, enabling always-on health monitoring without compromising comfort, security, or battery life. <iframe id="JotFormIFrame-250563172046452" src="https://form.jotform.com/250563172046452" frameborder="0" style="min-width:100%;max-width:100%;height:400px;border:none" scrolling="no">     </iframe> <h3 id="isPasted">In this webinar, we’ll explore:</h3><ul><li>The power-efficiency breakthrough that allows body-worn AI devices to run longer and process real-time health insights at the edge</li><li>Privacy-first AI innovations, including millimeter-wave radar and local data processing, that minimize reliance on cloud connectivity</li><li>Real-world applications of Ambiq’s technology, from fall detection and heart rate monitoring to advanced hearing aids and early dementia detection</li><li>What’s next for edge AI in digital health: how ultra-low-power AI is shaping the future of personalized care and smart health ecosystems</li></ul><h3 id="isPasted">Event details</h3>Webinar Date: Apr 8, 202 | 6 pm CET | 9:00 PSTWebinar Duration: 1 hour. 45 minutes discussion with 15 minutes for QnA. <h3>About the speakers:</h3>Carlos Morales, VP of AI at AmbiqCarlos Morales is the current Vice President of Artificial Intelligence (AI) at Ambiq, where he leverages the Subthreshold Power Optimized Technology (SPOT®) platform to advance AI across extensive mobile and portable devices within the IoT universe. His strategic focus includes pioneering low-latency, AI-driven applications while minimizing energy consumption by implementing local and intelligent analytics.Bringing over three decades of research and development experience from silicon to cloud, Morales has a distinguished career marked by significant technological innovations. Before his role at Ambiq, he spearheaded a comprehensive AI strategy and business development with Intel®'s global AI Software Team, where he served for five years. Earlier in his career, Morales was instrumental in Cisco’s Advanced Architecture and Research group, where he led initiatives such as the development of Fog Computing and the creation of the first full-featured software-defined router, now crucial in Edge to Cloud operations.Morales enthusiastically describes his decision to join Ambiq: “The moment I understood Ambiq’s unique value, I knew I had to be part of their journey. The opportunities unlocked by bringing AI to the vast far-edge ecosystem are mind-boggling and terrifically exciting. Ambiq is uniquely positioned to transform the computing landscape, and I am thrilled to contribute to this revolutionary path.”<h3>About the host: </h3>Samir Jaber is an editor, writer, and industry expert on topics of technology, science, and engineering. He is the editor-in-chief of the 2024 State of Edge AI and 2023 Edge AI Technology reports with Wevolver. Samir is the Chief Editor and Founder of Wryters, a content marketing and consulting agency. He has comprehensive experience working with Fortune 500 companies and industry leaders as a writer, editor, content manager, and consultant. He is an online content specialist with an academic background in mechanical engineering, nanotechnology, and scientific research. Samir is also a featured author in 30+ industrial magazines with a focus on Artificial Intelligence (AI), the Internet of Things (IoT), 3D printing, Autonomous Vehicles (AV), nanotechnology, materials science, and sustainability. His experience includes award-winning engineering research and patented engineering design in the fields of nanofabrication and microfluidics.

Register for our latest Edge AI webinar  |  Apr 8, 2025 06:00 PM CET. 

Powering the Future of Digital Health: Ultra-Low-Power AI for Smarter Patient Monitoring

healthcare

Autonomous Mobile Robots employed for material handling; Credits: Teradyne

This article offers an in-depth view of AMR robots, their underlying technology, and applications crucial in today's rapidly evolving technological landscape.

Understanding AMR Robots: A Comprehensive Guide

This is the 2nd article in a 3 part series featuring the applications of tinyML. The series introduces the concept of tinyML and explains some selected applications of the technology in various fields like sustainability, healthcare and education.By adopting digital advancements, healthcare providers are updating their services to improve them and make them accessible to more people. 5G networks, Augmented Reality (AR) &amp; Virtual Reality (VR), Wearables, Digital Twins, and many more technologies are enabling this digital transformation.One such technology with great potential in ensuring healthy lives and well-being for all is tinyML. The concept is a crossover between Artificial Intelligence (AI) and embedded systems that involves developing solutions that bring Machine Learning (ML) capabilities to edge devices. Benefits include privacy, affordability, low-power utilisation, the ability to work in remote locations, and a lot more.In our <a href="https://www.wevolver.com/article/tinyml" target="_blank">previous article</a> about the applications of tinyML, we explained how it unlocks new possibilities for sustainable development technologies. This article continues the series by showcasing two more applications of tinyML for medical research and improved healthcare.<h2 dir="ltr">MaRTiny: a tinyML solution for extreme heat sensing for urban climate science</h2>As a direct effect of global warming, summers are getting warmer. [1] This disruption of the natural balance poses a risk to all living entities on the planet. Higher temperatures are known to cause a degradation in the air quality and induce severe health-related issues like heat strokes. Looking at the data of weather-related fatalities in America, heat is the single most significant cause of weather-related deaths over the last 30 years, surpassing floods and storms. [2]But it’s not just the ambient temperature that causes discomfort. It’s the Mean Radiant Temperature (MRT) which accounts for the influence of all the surfaces in the surrounding that ultimately determines the occupant’s comfort. Often, the effect of roads, buildings and other entities has an adverse effect on the MRT, creating a complex problem intersecting health and urban infrastructure.To aid the collection of information related to the MRT, a tinyML solution called MaRTiny was developed by researchers from Arizona State University (ASU). It is a $200 low-cost mobile platform for bio-metrological sensing.<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjQyNTE3MzYxNzgyLTE2NDI1MTczNjE3ODIucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" width="720" height="297" id="isPasted" class="fr-fic fr-dii" alt="MaRTiny-tech-stack">The tech stack of the MaRTiny platformThe list of sensors embedded on the device include:<ul><li dir="ltr">Ambient temperature sensor</li><li dir="ltr">Globe temperature sensor for sensing the radiant heat</li><li dir="ltr">Humidity sensor</li><li dir="ltr">Ultra Violet (UV) sensor</li><li dir="ltr">Anemometer for measuring wind speed and direction.</li></ul>The platform uses an <a href="https://www.wevolver.com/profile/arduino.pro" target="_blank">Arduino</a> Uno to collect the data and a NodeMCU to provide it with the electronic infrastructure required for wireless data communication. At the core of the system is the Nvidia Jetson Nano development board that provides MaRTiny with the power to run tinyML algorithms.<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjQyNTE3NDcxMTAyLU5WSURJQV9KZXRzb25fTmFub19EZXZlbG9wZXJfS2l0Xyg0MDY1MDQyNTUwMykuanBnIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==" style="width: 300px;" class="fr-fic fr-dib" alt="nvidia-jetson-nano-development-board">The Nvidia Jetson Nano development board provides MaRTiny with image processing and ML capabilitiesJetson Nano, with its Camera Serial Interface (CSI), can connect with a camera for shadow detection, object detection and people-in-shade detection. Jetson’s ARM Cortex A57 MPcore, 128 <a href="https://www.wevolver.com/article/understanding-nvidia-cuda-cores-a-comprehensive-guide">CUDA core</a> Maxwell GPU coupled with 4 GB RAM is just enough to offer advanced ML capabilities on edge.The task of MRT estimation was formulated as a supervised learning problem by the inventors that required a lot of true data for training. This true data was collected with the help of a previously developed, larger and more expensive MRT detecting system going by the name of MaRTy. [3]&nbsp;<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjQyNTE3NTMzMzM5LTE1MDB4NTAwLmpwZyIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=" style="width: 300px;" class="fr-fic fr-dib" alt="MaRTiny-MaRTy">MaRTiny aims to offer similar functionality as that of MaRTy (in the image) at just 1% of the price. Image source: @ASUMaRTy, TwitterIt was observed that a Support Vector Machine (SVM) with a Radial Basis Function (RBF) kernel achieved the highest accuracy with the MaRTiny. The method being computationally light, offered an easy way of deployment on Jetson. To improve the estimations, both SVM and Artificial Neural Network (ANN) models were utilised and trained with data collected via sensors. [4]For privacy, the images taken by MaRTiny are not stored in the cloud or even in the device’s memory. It automatically discards personally identifiable information (PII) before reporting anything back to the servers, which is central to the idea of tinyML and a must-have feature for public deployment.<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjQyNTE3NTc1MjI0LWltYWdlNy5wbmciLCJlZGl0cyI6eyJyZXNpemUiOnsid2lkdGgiOjk1MCwiZml0IjoiY292ZXIifX19" style="width: 300px;" class="fr-fic fr-dib" alt="MaRTiny-setup">The experimental setup of MaRTinyThe tinyML solution, MaRTiny, can provide crucial insights related to weather conditions and the utilisation of shades and space by the public. Multiple MaRTiny devices can be installed across the city to gather valuable data for architecture and town planning.<h2 dir="ltr">Pneumonia Detection using Embedded Machine Learning</h2>Pneumonia is a respiratory infection caused by viruses, bacterias and fungi. According to the World Health Organisation (WHO) data, it is estimated that as much as 22% of all deaths in children aged from 1 to 5 is due to pneumonia. Bacteria induced pneumonia can be treated with antibiotics and other low-cost and low-tech medical facilities if detected at the right time. [5]Chest X-ray, blood oxygen levels and Complete Blood Count (CBC) are used to detect pneumonia which is time-consuming. Doctors manually review the chest X-ray and decide if the person is infected, which is sometimes inaccurate and requires an automated solution with better accuracy. This is the idea behind the tinyML solution that we are about to discuss in this section.Arijit Das, a 15-year-old tinyML enthusiast, has developed a tinyML model using the <a href="https://www.wevolver.com/profile/edge.impulse" target="_blank">Edge Impulse</a> platform to detect pneumonia from chest X-rays. The model can be deployed on single board development platforms like Raspberry Pi 4, Sony Supersense and a few other devices. With just the main board, a camera shield and the Edge Impulse tinyML model, the tiny setup can detect pneumonia in under a minute which is phenomenal considering the fact that it takes anywhere between 1 to 4 days for detection by manual inspection.<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjQyNTE3NzA0Nzk2LVJhc3BiZXJyeV9QaV80X01vZGVsX0JfLV9TaWRlLmpwZyIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=" style="width: 300px;" class="fr-fic fr-dib" alt="Raspberry-Pi-4">Credit card-sized single-board computers like Raspberry Pi come with enough compute power to run optimised tinyML algorithmsThe tinyML model doesn’t require internet connectivity to work and offers an accuracy of over 80%, which is far better compared to manual inspection. This can prove to be quite helpful in largely affected regions like Sub-Saharan Africa and South Asia, where a huge number of examinations need to be conducted related to the same disease.&nbsp;At the time of writing this article, the project is fully functional and has been clinically tested by over 10 doctors in a local survey conducted by the inventor. The project is available on <a href="https://github.com/Pneumonia-Detection-using-EdgeML" target="_blank">GitHub</a> for everyone to try, test and contribute. [6]<h2 dir="ltr">Conclusion</h2>It’s the third year into the pandemic, and the emphasis on healthcare research is more than ever. Introducing technologies that can work with little power and no network connectivity at an affordable price in healthcare can save several lives. Leveraging interdisciplinary technologies like tinyML is the way forward to a safer and healthier future.About tinyML FoundationWith its workshops, webinars, expert talks, research conferences, competitions, and a plethora of other events, tinyML Foundation enables students, researchers, and industry personnel to come together and share their experiences about the technology. It is the incredibly collaborative nature of the community that allows it to advance rapidly.<img src="https://images.wevolver.com/eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNjQyNTE3NzY5MDkxLWV5SmlkV05yWlhRaU9pSjNaWFp2YkhabGNpMXdjbTlxWldOMExXbHRZV2RsY3lJc0ltdGxlU0k2SW1aeWIyRnNZUzh4TmpNMk9UZ3lNVEUwT1RRNExWUnBibmxOVEdadmNrZHZiMlFnS0RRd01EQWdlQ0EzTlRBZ2NIZ3BMbXB3WnlJc0ltVmthWFJ6SWpwN0luSmxjMmw2WlNJNmV5SjNhV1IwYUNJNk9UVXdMQ0ptYVhRaU9pSmpiM1psY2lKOWZYMD0ud2VicCIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=" style="width: 300px;" class="fr-fic fr-dib" alt="tinyML-banner"> The <a href="https://www.wevolver.com/article/tinyml-for-good-where-machine-learning-meets-edge-computing" target="_blank">introductory article</a> familiarised the readers with the concept of tinyML.The&nbsp;<a href="https://www.wevolver.com/article/tinyml" target="_blank">first article</a> explained some possibilities of tinyML applications in sustainable technologies.The <a href="https://www.wevolver.com/article/applications-of-tinyml-in-healthcare-for-a-safer-future/" rel="noopener noreferrer" target="_blank">second article</a> is about tinyML applications aiding healthcare and medical research.The third and final article shall cover the tinyML applications in&nbsp;pedagogy&nbsp;and education.<hr><h2>References</h2>[1] Extreme Weather Toolkits, Climate Central, [Online], Available from: &nbsp;<a href="https://medialibrary.climatecentral.org/extreme-weather-toolkits/extreme-heat" target="_blank">https://medialibrary.climatecentral.org/extreme-weather-toolkits/extreme-heat</a>[2] Weather Related Fatality and Injury Statistics, National Weather Service, [Online], Available from:&nbsp;<a href="https://www.weather.gov/hazstat/" target="_blank">https://www.weather.gov/hazstat/</a>[3] Ariane Middel, E. Scott Krayenhoff, ‘Micrometeorological determinants of pedestrian thermal exposure during record-breaking heat in Tempe, Arizona: Introducing the MaRTy observational platform’, Science of The Total Environment, Volume 687, 2019, Pages 137-151, ISSN 0048-9697, Available from:&nbsp;<a href="https://doi.org/10.1016/j.scitotenv.2019.06.085" target="_blank">https://doi.org/10.1016/j.scitotenv.2019.06.085</a>[4] Karthik K. Kulkarni, 2021, ‘Machine Learning and Vision Using Edge Devices for Multimodal Chatbots andBio-meteorological Sensing’, MS Thesis, Arizona State University, United States of America, [Online], Available from:&nbsp;<a href="https://keep.lib.asu.edu/_flysystem/fedora/c7/Kulkarni_asu_0010N_21269.pdf" target="_blank" data-lf-fd-inspected-kn9eq4roawkarlvp="true">https://keep.lib.asu.edu/_flysystem/fedora/c7/Kulkarni_asu_0010N_21269.pdf</a>[5] Pneumonia Facts Sheet, World Health Organisation, [Online], Available from:&nbsp;<a href="https://www.who.int/news-room/fact-sheets/detail/pneumonia" target="_blank">https://www.who.int/news-room/fact-sheets/detail/pneumonia</a>[6] Arijit Das, Pneumonia-Detection-using-EdgeML, GitHub, [Online], Available from:&nbsp;<a href="https://github.com/Pneumonia-Detection-using-EdgeML" target="_blank">https://github.com/Pneumonia-Detection-using-EdgeML</a>

In this article, we take a look at two tinyML projects related to healthcare. The first project helps gather Mean Radiant Temperature data outdoors to protect people from extreme heat, and the other one is a solution for affordable, accurate, & rapid detection of pneumonia.

healthcare

Technical Guides

Understanding AMR Robots: A Comprehensive Guide

Applications of TinyML in healthcare for a safer future

Profiles