MedTrack

A-powered pharmaceutical intelligence for medicine quality, supply availability, and diversion detection.

The problem

Health systems lose medicines at every stage of the supply chain, and most of that loss is invisible until a clinic shelf is already empty. Each module below is grounded in documented field research rather than assumption:

• Quality — substandard and falsified medicines remain a persistent, measurable risk in low- and middle-income health systems, as documented by the WHO and peer-reviewed literature (BMC, PLOS).
• Availability — facility-level stockouts are common even when national-level procurement looks adequate on paper. The gap is visibility, not just supply.
• Diversion — donated and government-purchased medicines disappear before reaching patients at meaningful scale. Field studies in Malawi found 30–35% of government-purchased drugs went missing before reaching clinics; Togo lost over $1M in Global Fund-supplied antimalarials to theft.

MedTrack treats these as one connected problem instead of three separate ones, because in practice they are. A stockout, a diversion event, and a quality failure can look identical from a clinic's point of view until there's data to tell them apart.

What it does

Module	Function
Quality Verification	Verifies medicine authenticity and quality at point of dispensing
Availability / Forecasting	Predicts stockout risk from supply-chain event history
Diversion Detection	Flags abnormal loss by reconciling received vs. dispensed volumes

All three modules read and write a shared stock_events data model, so one pipeline of medicine-movement data powers quality checks, supply forecasts, and diversion alerts together — rather than three disconnected tools.

Tech stack

• Backend: Python 3 / FastAPI
• Data: SQLAlchemy ORM, SQLite for pilot deployments (Postgres-ready for scale)
• Design principle: country-agnostic data model from day one

Project status

Early-stage. PRD, architecture documentation, and business case are complete. The repo scaffold is in progress — the original quality-verification logic is being merged in, and the dataset acquisition plan is being finalized in parallel.

Project structure

medtrack/
├── app/
│   ├── main.py                       # FastAPI entrypoint
│   ├── database.py                   # SQLAlchemy engine/session
│   ├── models/
│   │   └── stock_event.py            # shared stock_events model
│   ├── modules/
│   │   ├── quality/verification.py   # quality verification (stub — merge in progress)
│   │   ├── availability/forecasting.py
│   │   └── diversion/reconciliation.py
│   └── api/
│       └── dashboard.py              # combined dashboard endpoint
├── tests/
│   └── test_smoke.py
├── docs/
│   ├── PRD.md
│   ├── architecture.md
│   └── business-case.md
├── requirements.txt
└── .env.example

Getting started

Run one of the setup scripts to scaffold the project and install dependencies:

• setup-termux.sh — Android / Termux
• setup-vscode.sh — macOS / Linux desktop

Then, from the project root:

uvicorn app.main:app --reload

License

MedTrack is proprietary, commercially licensed software — see LICENSE. For pilot, partnership, or commercial licensing inquiries, contact PharmaLens, Uganda