Team Tuan and Evan

oculusMD

Your AI copilot that monitors medication routines and guides proper consumption on the patient’s behalf.

oculusMD

Video Demo

About this project

oculusMD is an AI-powered medication safety assistant designed to protect patients after they leave the hospital — the moment when medication errors are most common and most dangerous. Every year, millions of patients are discharged with complex multi-drug regimens and dense paperwork they struggle to understand. oculusMD bridges that gap by putting a real-time clinical safety layer directly on the patient's phone. How it works: Point the camera at any medication bottle or pill packet. oculusMD uses Google Gemini 2.5 Flash to instantly identify the drug from the label, then runs it through a safety engine backed by a structured drug interaction database in MongoDB. In seconds, the app checks three things: Contraindications — Is this medication dangerous given the patient's medical history? (e.g., a decongestant for a patient with severe hypertension) Drug-drug interactions — Do any of the medications being taken together create a dangerous or lethal combination? (e.g., Warfarin + Metronidazole, a combination that can cause fatal bleeding) Too-soon re-dosing — Has this medication already been taken recently? The app tracks every confirmed dose and blocks an accidental double-dose before it happens. Every scan result is logged to the Solana blockchain, creating a tamper-proof audit trail that patients and caregivers can share with their care team. For discharge documents: Patients also struggle to understand the dense medical language in discharge paperwork. oculusMD lets patients photograph their discharge documents — even multi-page packets — and uses Gemini to extract and translate the instructions into plain English. The summary is read aloud in a natural voice via ElevenLabs TTS. Patients can also ask follow-up questions in plain language ("What medications do I take at home?", "Can I drink alcohol?") and get spoken, grounded answers pulled directly from their own documents. Tech stack: React Native (Expo), FastAPI, MongoDB, Google Gemini 2.5 Flash, Eleven

Gallery