Inspiration
People with health conditions struggle to quickly identify appropriate meals and exercise routines that accommodate their medical restrictions, mental health needs, and physical limitations.
What it does
Vital Loop meets this crisis with a systems-level solution:
- Pulls medical history from MyChart to account for allergies, medications, and chronic conditions
- Scans lab results to detect nutrient deficiencies (iron, vitamin D, omega-3s)
- Syncs with Apple Health/Samsung Health for real-time food and hydration tracking
- Integrates with period-tracking apps like Flo to tailor meal plans around hormonal changes
- Offers 24/7 support via a personalized AI wellness coach trained on my full health profile Vital Loop then transforms this data into grocery-ready shopping lists and fitness routines that aim to help users achieve their defined goals.
How we built it
We started with a React-based web app that integrates multiple health data streams through a modular connector system - pulling from MyChart via FHIR APIs, syncing with Apple Health and Samsung Health for real-time tracking, and connecting to period-tracking apps like Flo. The core AI engine uses machine learning to synthesize medical history, lab results, hormonal cycles, and lifestyle data into personalized meal and fitness recommendations. Our breakthrough feature is the on-device OCR system that processes uploaded lab results locally, extracts key health markers like vitamin D levels, asks for user confirmation, then securely deletes the original image - ensuring privacy while enabling deep personalization. We built the conversational AI coach using natural language processing trained on evidence-based nutrition and wellness research, with real-time adaptation that learns from user preferences and adjusts recommendations instantly when they decline suggestions.
Challenges we ran into
Getting different health APIs to play nice together was way harder than expected - MyChart uses FHIR, Apple Health has its own format, and Flo tracks periods completely differently. We also faced a major privacy paradox: we needed deep personal data to give good recommendations, but users rightfully freaked out about sharing medical records. Our solution was building on-device OCR that extracts lab results, asks for confirmation, then deletes the original image, but getting that secure processing right took forever.
Accomplishments that we're proud of
We built something that actually learns and adapts in real-time - when our demo user Sarah declined the complex quinoa bowl and instantly got a simple sandwich suggestion, that felt like magic. More importantly, we created an app that could have genuinely helped Ashanti during her struggles at Cornell, addressing a real mental health crisis affecting 70% of college students. Our technical breakthrough with secure, on-device OCR processing means users can upload bloodwork, get insights about vitamin deficiencies, and know their sensitive data never leaves their phone.
What we learned
- Users Want Control, Not Perfection: People don't want another rigid meal plan - they want options and the ability to say "not today." Building in decline buttons and modifications was crucial.
- Data is Only as Good as Context: Raw numbers don't help anyone. Connecting "low iron + feeling tired + heavy periods" into actionable recommendations is where the real value lives.
- Mental Health is Physical Health: The research was eye-opening - 60% of American calories come from ultra-processed foods, and there's a direct link to depression rates. Food isn't just fuel, it's medicine.
What's next for Vital Loop
We're planning 12-week trials with student health centers to prove our approach works, targeting college students who are struggling with mental health at unprecedented rates as our perfect early adopters. Our roadmap includes partnering with researchers for peer-reviewed studies, rolling out features gradually so users can opt into complexity as they see value, and long-term integration with insurance plans as preventive care. The vision is making personalized, evidence-based wellness as normal as checking your weather app.
Built With
- note.js
- openai
- tensorflow.js



Log in or sign up for Devpost to join the conversation.