Inspiration
For elderly individuals, patients managing chronic conditions, and busy careers, keeping track of daily medication is a constant source of stress. Missing a dose or running out of critical pills can lead to serious health complications. While smartphones can help, existing solutions are often overly complex, require a constant internet connection, or push unsafe AI medical advice. We wanted to build a reliable, offline-first mobile assistant that keeps daily routines strictly organized, ensures users never run out of medication, and utilizes AI purely as a safe, strictly fenced information tool rather than an unverified pseudo-doctor.
What it does
Dashboard & Daily Schedule: Displays today’s medication overview with exact-time alarms powered by Android’s native AlarmManager, ensuring notifications fire reliably even after device reboots. Smart Supply Tracking: Automatically calculates pill depletion based on daily usage and alerts the user with a "Time to Buy" notification 3 days before running out. Prescription Scanning (OCR): Uses image recognition to turn printed prescription photos into text, featuring a dedicated consent check and explicit confirmation prompt before data can be used. Safety-Fenced AI Helper: Integrates Google's Gemini API for general medicine inquiries, guarded by multi-layer safety mechanisms. Offline Emergency Screen: Quick access to essential storage guidelines and local Sri Lankan emergency hotlines (1990 Suwa Seriya).
How we built it
Used Python-Kivy to build Frontend and UI and used SQLite to store user's information (Name and Birthday), medicine information, schedule information and pill depletion. As well as, Added Gemini ai assistance to get medicine details. Gemini model is connected to our project by an API key which is in
Challenges we ran into
Background processes on modern mobile OSs frequently kill scheduled tasks. Ensuring alarms fire at exact times and persist across device restarts required writing custom Java bridging code and hooking into system-level broadcast receivers and build ai assistance.
Accomplishments that we're proud of
Developed an Android-compatible medication management application using Python and Kivy. Integrated prescription scanning using OCR technology. Built a safety-first AI assistant powered by Google's Gemini API.
What we learned
How to properly interface native Android Java classes directly from a Python codebase using PyJNIus. Optimizing SQLite query flows to support quick mobile UI re-renders and reliable state tracking for medication stock.
What's next for Smart Medication Assistant
Move prescription text recognition fully offline using local mobile OCR libraries. Allow family members or caregivers to receive remote notifications if an elderly user skips a critical dose.