HealthMobi.
A medication-adherence ecosystem that closes the loop between reminder, physical intake, and long-term progress — with an AI assistant that takes natural-language schedule changes.
Course → reminder → confirmed intake → adherence matrix, across app and pillbox.
Illustrative workflow · explore the engineering above
The problem
Medication adherence fails quietly. A reminder fires, but nothing confirms the dose was actually taken; schedule changes mean tapping through forms; and progress is invisible until a check-up. Patients need the reminder, the confirmation, and the insight to work as one loop.
What I built
Built phone-OTP authentication and medication courses with customizable schedules. Cron jobs drive reminders through push notifications and SMS, and each reminder can be confirmed with a single press — or automatically by the smart pillbox reporting physical intake. Every confirmation feeds a visual adherence matrix. An AI assistant accepts natural-language commands like “move breakfast to 8 AM” to adjust meal times and schedules, extracts structured data from prescriptions, and uses a Chroma vector store for context. WebSocket updates keep app and pillbox in sync in real time.
The engineering thinking
Adherence improves when the reminder, the confirmation, and the progress view form one loop instead of three disconnected features. The AI assistant exists because schedule changes are the moment patients usually give up — natural language removes the friction. The pillbox closes the gap between a digital confirmation and a physical dose actually being taken.
The outcome
A working ecosystem spanning the Flutter app, the Node.js/Express backend with MySQL and WebSockets, the AI layer, and the embedded pillbox. Daily motivational quotes, custom notes, and adherence insights round out the patient experience. An open-source release is planned to make the stack available for broader healthcare experimentation.