Parkinsonvalesce

AI-driven Pakinson therapy through playing games. Clinical tapping biomarkers meet nature-inspired exercises to track, monitor, and brighten the Parkinson’s journey. Play, track and convalescence.

Comment

Inspiration: Turning Powerlessness into Action

A few years ago, our father was diagnosed with Parkinson’s Disease (PD). As non-medical professionals, my sister and I felt a profound sense of powerlessness watching him struggle with tasks that were once second nature. We learned that while Parkinson’s is currently incurable, clinical research proves that targeted exercise can significantly slow motor deterioration. We discovered that finger dexterity is a critical biomarker for PD progression. We wanted to build a tool that could transform our father's daily "therapy" from a chore into a moment of light and engagement. Our mission was to create an app that allows patients to train their fine motor skills while providing a data-driven way to monitor their condition from the comfort of home.

What it does: What is MorningStar

Morning Star is a therapeutic mobile application designed to help Parkinson’s Disease (PD) patients maintain motor control through nature-inspired daily exercises. Unlike clinical tests that feel cold and intimidating, Morning Star transforms therapy into a "Morning Flow"—a suite of games like Sun Chaser and Bloom Pulse that gamify essential motor movements. The app serves two purposes: Therapeutic Exercise: Games are specifically designed to target bradykinesia (slowness), hypokinesia (small movements), and tremors. Clinical Monitoring: While the user plays, the app calculates digital biomarkers, such as Mean Dwell Time and Inter-Tap Interval variability, providing a data-driven report that tracks disease progression over time.

How we built it: From Research to Deployment

Our development workflow was an iterative loop between the Chat Prompt and the Build Tab in Google AI Studio: Research & Logic: We used the Chat Prompt with Google search grounding to synthesize medical papers on bradykinesia and "Sequence Effect". We defined our core metrics: Mean Dwell Time and Velocity Decrement. Visual Identity: We moved to the Build Tab in AI Studio to establish the "Parkinsonvalesce" aesthetic—nature-inspired, warm, and high-contrast for elderly accessibility. Prototyping: We utilized Gemini to generate a high-performance React frontend. We implemented a "Signal Cleaning Layer" to distinguish between voluntary taps and involuntary tremors. Using the Thinking model to calculate user's level. Use the Gemini API to search and summarize academic paper. Validation: We pushed our code back into the Chat Prompt for "stress testing," asking Gemini to simulate potential bugs or accessibility hurdles for PD patients.

Challenges we ran into: Overcoming the Technical Gap

The journey was not without its hurdles: The Language Barrier: This was our very first React application. We had zero knowledge of TypeScript or the React lifecycle. When the build failed, we didn't know why. However, Gemini acted as a 24/7 tutor, explaining bugs in "human language" and teaching us how to fix them in real-time. *Time Compression: In a traditional setting, conducting medical research and building a validated prototype would take months. As non-experts, we faced a steep mountain of data. Gemini compressed this timeline, allowing us to summarize complex clinical papers and deploy a functional, research-backed app in under one week.

Accomplishments that we're proud of

Morning Star is a working app which is ready to go. It is more than just an entry for a hackathon; it is our way of telling our father—and the 10 million others living with Parkinson's—that they are not alone, and that technology can be a source of hope and strength. Morning Star proves that with the right AI tools, a personal story can be transformed into a functional tool for health and hope.

What we learned: The Era of "Vibe Coding"

As beginner developers, the most transformative lesson was discovering the power of Google AI Studio and the Gemini API. We pioneered a process we call "Vibe Coding"—using natural language to bridge the gap between medical intent and technical execution. Key takeaways included: The "AI Project Team" Although our human team only has two members, Gemini acted as a full department. By utilizing System Instructions, we assigned Gemini specific personas: a Medical Researcher to verify biomarkers, a Senior React Developer to structure the code, and an Occupational Therapist to design the user experience. Adaptive Difficulty Logic We learned to implement dynamic scaling. The app doesn't just give a flat score; it uses generative logic to define levels based on user performance. For example, we calculate the Coefficient of Variation (the standard deviation and the mean of the Inter-Tap Intervals) for tap intervals to adjust game speed.

What's next for Morning Star

To move Morning Star forward, we plan to evolve the prototype into a comprehensive multimodal health suite. This includes integrating voice and motion sensors to track speech clarity and tremors, alongside a clinical reporting dashboard that correlates motor performance with medication cycles for neurologists. We aim to personalize the experience through adaptive AI routines that adjust difficulty based on daily symptom fluctuations and foster emotional support via a caregiver "Circle of Care" system. Ultimately, our goal is to conduct a clinical pilot, transforming a nature-inspired game into a research-backed instrument that improves autonomy and slows progression for the Parkinson's community.

Built With

Share this project:

Updates