Explore a glimpse into the future of medicine through the latest research from Hopkins BME. This page provides a curated selection of our most recent publications, spanning a wide range of topics and showcasing how our faculty and students are turning bold ideas into real-world solutions. From engineering transformative new treatments to pioneering the next generation of medical devices, our community is committed to creating tangible breakthroughs and pushing the boundaries of what’s possible.
A New Clue for Balance Therapy
Treating Superior Canal Dehiscence Syndrome (SCDS)—an inner ear disorder that triggers constant dizziness and balance issues—has always been a challenge. But now, Kathleen Cullen and her team have conducted the first study that used wearable motion sensors to objectively measure how SCDS affects daily life. They discovered that SCDS patients unconsciously adopt a clever trick to cope: holding their heads rigidly still when navigating complex environments like walking in the dark or on stairs. The researchers suggest this reduced head motion is a protective adaptation aimed at minimizing stimulation of the affected inner ear canal. The study provides the first quantitative roadmap linking the inner ear problem to real-world movement, setting the stage for more targeted and effective balance therapy to help patients regain confidence and stability in their daily lives.
A New Era for High-Speed Eye Disease Screening
Johns Hopkins engineers, led by Nicholas Durr, have built a state-of-the-art eye camera that makes diagnosing serious eye diseases like glaucoma easier and faster. Current eye cameras require perfect focus, which is difficult if a patient has strong prescriptions or poor vision. This new device solves that by taking a picture of the fundus—the back interior surface of the eye, which contains the retina—using a special digital lens. The camera captures data that can be digitally refocused after the picture is taken—it can “fix” blurry images for people with severe nearsightedness or farsightedness. This new approach allows doctors to get crystal-clear images of the retina every time, greatly improving the speed and reliability of screenings for conditions like glaucoma.
Saving Time for Genetic Counselors
The growing demand for genetic testing often forces Genetic Counselors (GCs) to spend valuable time on administrative tasks instead of patient care. To tackle this challenge, a collaborative study led by Casey Taylor unveiled the potential of natural language processing (NLP) to precisely track time spent in counseling sessions across diverse clinical specialties, with the core goal of optimizing the time GCs can devote to patients. “This study not only sheds light on real-world evidence of time spent in genetic counseling visits, it provides a strategy for healthcare providers to understand, and ultimately improve their processes,” says co-author and genetic counselor Carolyn Applegate. By applying advanced models to a substantial, seven-year dataset of genetic counseling notes, the researchers found that the median time spent in a session was 50 minutes, with noticeable variations based on clinical specialty, time periods (pre-COVID and during the COVID pandemic), delivery modes (in-person and telehealth) and phases of counseling (pre- and post- genetic testing). Promising methods developed from this study are being validated in the Sequence of Cardiovascular Genetic Counseling and Testing clinical trial led by Cynthia James (RESEQUENCE-GC, NCT05422573).
Journal of the American Medical Informatics Association | November 2025
Engineering Living Implants for the Brain
The biggest obstacle to creating reliable, long-term brain-machine implants is the body’s rejection of the stiff, foreign material, which causes the device to fail. In a new perspective paper, Johns Hopkins researcher Xiao Yang and student Junpeng Li (MSE ’26) chart the future of neurotechnology designed to eliminate this rejection problem. Made of flexible and stretchable electronics that mimic soft brain tissue, these devices are designed to co-evolve and regenerate with the body, enabling the implant to last a long time. By combining this seamless integration with technologies like AI for real-time decoding and organoid intelligence (OI) for hybrid computing, these advanced neurotechnology devices will be far more than passive sensors. They are envisioned as active, living, and learning extensions of the nervous system, promising long-lasting and effective treatments for countless neurological conditions.
New Tech for Safer Cardiac Drugs
While miniature, lab-grown hearts called cardiac organoids are powerful tools for modeling disease, traditional recording methods only capture their electrical signals in 2D. This limitation means researchers miss vital information about how signals truly travel in the human heart. To overcome this, Deok-Ho Kim and a multidisciplinary team have developed a novel, shape-adaptive shell microelectrode array (MEA) that fully wraps the organoid. This Shell MEA technology enables comprehensive 3D electrophysiological mapping, creating high-resolution maps of the heart’s electrical pathways. The platform transforms static organoids into dynamic, functional models that can more accurately predict a human heart’s reaction to new drugs. This capability will significantly speed up drug discovery and help advance personalized medicine, ensuring cardiovascular treatments are safer and more effective for individual patients.
Inside the Brain’s Sound System
How does the brain process all the sounds we hear, from a single musical note to the roar of a concert crowd? Patrick Kanold and colleagues have delved into the brain’s “wiring” for sound and found that it is far more precise and organized than previously thought. Using advanced brain imaging, the team found that our brains actively “connect the dots” between different tones to help us interpret complex sounds like speech and music. The research gives new insight into how our brains decipher the world of sound and may inform future studies on hearing disorders.
A Robotic Touch for More Reliable Diagnosis
Manual palpation, a technique used to feel for abnormalities in tissues, can be inconsistent due to variations in a clinician’s speed and experience. Researchers led by Nitish Thakor developed an automated robotic palpation system that uses a special tactile sensor that mimics human touch and accurately detects different fracture types regardless of scanning speed. In tests, the device identified fractures in a chicken wing model with 99.8% accuracy, showing potential to provide a more reliable and objective way to diagnose conditions in both hard and soft tissues.
IEEE Transactions on Medical Robotics and Bionics | May 2025
The Sink-Index: Finding Answers for Dementia
Frontotemporal dementia is a devastating illness that is notoriously difficult to diagnose, often mistaken for Alzheimer’s disease in its early stages. A team led by Sri Sarma and Chiadi Onyike has developed a promising new diagnostic tool using routine electroencephalogram (EEG) readings of brain waves. Called the “Sink-Index,” their approach analyzes the brain’s electrical activity in a new way, revealing a unique pattern that accurately distinguishes between patients with frontotemporal dementia, those with Alzheimer’s disease, and healthy controls. This discovery paves the way for a future where doctors can use a non-invasive EEG to more accurately diagnose complex forms of dementia and begin treatment sooner.
Smarter Spine Surgery with AI
Researchers at the Imaging for Surgery, Therapy and Radiology (I-STAR) Lab developed an innovative method to improve surgical precision using machine learning and ultrasound. Their new framework trains an AI model to accurately identify and track anatomical structures by automatically transferring labels from 3D diagnostic images to live 2D ultrasound scans. When applied to spinal surgery, the method precisely segmented and tracked vertebrae with minimal error, demonstrating its potential as a tool for real-time surgical navigation.
A Life-Saving Alarm to Protect Premature Infants
Bubble continuous positive airway pressure (bCPAP) is a vital treatment for newborns with respiratory distress, but a simple disconnection in the system can go unnoticed in a busy Neonatal Intensive Care Unit (NICU), leading to dangerous drops in oxygen. To solve this problem, students and faculty in the Undergraduate Design Team Program developed an automated monitoring device that sounds an alarm if the bubbling stops, alerting nurses within seconds. In initial tests, the device detected disconnections in just over four seconds.
Conquering Disease with Genetic Data
What if we could create a highly accurate map of our genes to fight diseases? Using a massive dataset of genetic information, Alexis Battle and her team built “gene coexpression networks” that offer a clear picture of how genes work together to maintain our health and what goes wrong when we get sick. These maps help scientists pinpoint the genes responsible for diseases like cancer, potentially leading to better diagnostics and personalized treatments.
The Future of Personalized Immunity
In a new perspective paper, Derek Cummings and collaborators explain that an individual’s immune system is determined by three things: your genes, the unique changes your immune cells make throughout your life, and how your body responds to its environment. By continuing to explore and map these three factors across diverse populations, the researchers argue that medicine can finally move past the current one-size-fits-all approach and towards truly personalized healthcare. Unlocking this knowledge will pave the way for treatments that are precisely tailored to an individual’s immune system.
Analyzing Heart Shape to Predict Stroke Risk
The shape of the heart’s left atrial appendage (LAA) —a tiny pouch in the heart— is closely linked to a patient’s risk of stroke. Doctors currently lack a standard, objective way to measure and categorize the LAA shape, making it difficult to accurately predict a patient’s risk based on this vital clue. To solve this, Natalia Trayanova’s team built a new computational framework that combines elastic shape analysis with unsupervised machine learning to reliably categorize LAA morphology into robust shape clusters. By identifying specific high-risk shapes that are more likely to create blood clots, the system provides a more precise and consistent way to understand the link between LAA shape and stroke risk, paving the way for more accurate stroke prevention tools in the clinic.