H2AI 2025
From March 14–16, Georgetown University hosted the Healthcare Hackathon with AI (H2AI), a 36-hour event uniting clinicians, engineers, business leaders, and students with a shared goal: to build innovative prototypes at the intersection of AI and medicine. Drawing 240 participants from 20 institutions, the hackathon aimed to spark ideas that could transform patient care and clinical outcomes.
Team Rabbit, composed of students from MIT, George Mason, and UPenn, stole the spotlight. They not only took home the $1,000 cash prize sponsored by the Patient Safety Technology Challenge but also claimed the $5,000 Grand Prize of the competition. Their project, a VR-based AI platform focused on Parkinson’s disease, was praised by judges for its clinical validity, patient-centered approach, business feasibility, AI safety, and technical rigor.
From left to right, Team Rabbit Team Members: Yufei Chen, Zamir Osmanzai, Mohamed Shaik, Udit Verma, Edward Sun (photo provided by Georgetown University)
Team Rabbit’s platform simulates real-world mobility challenges—like navigating narrow doorways or busy corridors—using immersive virtual reality. These simulated environments allow the system to capture detailed neuromuscular and gait data, giving clinicians and caregivers continuous, objective insights into a patient’s motor function. By tracking how mobility changes over time in response to medication adjustments or disease progression, the platform aims to reduce fall risk and enhance quality of life for patients with Parkinson’s.
At the heart of Team Rabbit’s solution are two machine learning models designed to analyze patient data. The first model classifies Parkinson’s severity on a scale from 0 to 4 based on common drawing tasks like spiral and meander patterns. It uses image processing to isolate handwriting traces and employs Support Vector Machines (SVM) and Logistic Regression to distinguish identify individuals with Parkinson’s. The team’s second model, the Voice Severity Classification Model, is a hybrid model that uses both traditional machine learning and deep learning techniques to analyze vocal patterns, providing another layer of diagnostic insight into disease severity. In conjunction, these models offer a powerful toolkit for clinicians, enabling data-driven decisions and personalized care strategies.
For one member of Team Rabbit, the inspiration behind the project was personal. After a family friend suffered a hip injury and faced a long, difficult recovery, their home had to be completely modified to prevent future falls. “The family had to change the layout of the entire house,” the team member shared, “adding several preventative measures to ensure she did not fall again.” Given the higher risk of falls for individuals with Parkinson’s disease, that experience planted the seed for a solution that could help others minimize related challenges.
Team Rabbit is now working on a minimum viable product and plans to begin trials within the next year. Their roadmap includes developing user-friendly interfaces for both patients and providers, in close collaboration with Parkinson’s specialists and care teams.
As the healthcare industry continues to explore AI’s potential, innovations like Team Rabbit’s show just how powerful cross-disciplinary collaboration can be. With the right support and continued development, their platform could redefine how we monitor, manage, and ultimately improve mobility outcomes for patients with Parkinson’s disease.
