Aleph Hackathon

MediChain — Your health. Your data.

MediChain is a privacy-first medical records application that gives patients control over how their laboratory data is processed, organized, and shared. Instead of sending sensitive medical documents to a cloud AI provider, MediChain uses Tether QVAC for local AI processing.

The Problem

Medical records contain highly sensitive information, yet patients often need to upload, process, and share these documents across different systems. Using cloud AI to extract or analyze medical documents can mean sending sensitive data to external infrastructure, and patients may want to share only specific records with a doctor instead of exposing their entire medical history.

Our Solution

MediChain combines local AI processing with patient-controlled sharing. Here's how it works for the patient:

1. Sign in to their private account
2. Upload a laboratory report as an image
3. Process the document locally using QVAC OCR and local inference
4. Review the extracted laboratory results before saving anything
5. Confirm and store structured medical records
6. View their medical history and timeline
7. Select which records they want to share
8. Generate a Health Pass with a QR code or shareable link

A doctor can then open the Health Pass and access only the records the patient explicitly shared.

QVAC Integration

QVAC is a core part of MediChain's medical-document workflow. The local pipeline looks like this:

Laboratory image → QVAC OCR → local structured extraction → validation → patient review → confirmed medical records

MediChain also uses local QVAC inference to organize confirmed medical information and power AI-assisted features, without relying on a cloud LLM for the medical AI workflow. The extraction pipeline includes validation and grounding safeguards designed to prevent unsupported information from being silently added to a patient's records. The patient is always the final verification step before extracted results are stored.

Privacy by Design

MediChain keeps local AI processing separate from application persistence. QVAC handles the medical-document AI processing locally, while Supabase takes care of authentication and persistence. Patient and doctor accounts have separate roles, Row Level Security restricts patients to their own records, and doctors never get general access to patient data. Instead, sharing happens through a scoped Health Pass generated by the patient.

Human-in-the-Loop

MediChain never treats AI-extracted information as medical truth automatically. After QVAC processes a document, the patient gets a review screen where the extracted laboratory values can be checked against the original source before confirming. Only confirmed records make it into the patient's medical history.

Built With

Tether QVAC, React, TypeScript, Vite, Supabase, PostgreSQL + Row Level Security, Capacitor, Android

Demo Flow

Patient uploads a laboratory report → QVAC processes it locally → patient reviews extracted values → results are confirmed → medical history updates → patient selects records to share → MediChain generates a Health Pass / QR → doctor opens the Health Pass → only the selected records are displayed

Disclaimer

MediChain is a hackathon prototype built for organizing and selectively sharing medical records. It is not a medical diagnostic system and does not replace professional medical advice.