Medication Reconciliation Service

A backend service for ingesting patient medication data from multiple sources, detecting conflicts, and managing their resolution.

---

🚀 Live Demo

API Documentation (Swagger UI): 👉 https://medication-reconciliation-service-1.onrender.com/docs

---

⚙️ Setup Instructions (Run Locally)

# Clone the repository
git clone <your-repo-link>
cd medication-reconciliation-service

# Create virtual environment
python -m venv venv
venv\Scripts\activate  # Windows

# Install dependencies
pip install -r requirements.txt

# Set environment variable
set MONGO_URL=your_mongodb_atlas_url

# Run the server
uvicorn app.main:app --reload

---

Architecture Overview

The service follows a layered architecture:

• Routes → API endpoints (FastAPI)
• Services → Business logic (ingestion, conflict detection)
• Repositories → Database interaction (MongoDB)
• Models/Schemas → Data validation (Pydantic)

Flow:

1. Client sends medication data
2. Service compares with latest snapshot
3. Conflicts are detected
4. New snapshot stored with versioning
5. Conflicts stored separately for tracking

---

Assumptions & Trade-offs

• Conflict detection is based on simple field comparison (e.g., dose mismatch)
• Versioning is linear (no branching/merging of histories)
• No authentication added to keep scope focused
• MongoDB chosen for flexibility with nested medication data

---

Known Limitations & Future Improvements

• No user authentication / authorization
• Conflict resolution is manual only
• No pagination for reports
• No advanced conflict rules (e.g., drug interactions)

Next Steps:

• Add auth (JWT)
• Improve conflict detection rules
• Add audit logs
• Deploy with Docker for portability

---

Seed Data

You can populate sample data using:

python scripts/seed.py

---

Tests

Run tests with:

pytest

Covers core ingestion and conflict detection logic.

---

AI Usage

AI tools were used to:

• Speed up boilerplate setup (FastAPI structure, initial scaffolding)
• Assist in refining architecture and edge-case handling

Manual Work:

• Designed overall architecture and data flow
• Implemented conflict detection and versioning logic
• Structured repositories and services cleanly

---

Tech Stack

• FastAPI
• MongoDB (Atlas)
• Motor (async MongoDB driver)
• Pydantic
• Pytest

---

Deployment

• Hosted on Render
• Uses MongoDB Atlas for cloud database
• Environment-based configuration for production readiness

---

Key Features

• Medication ingestion with versioning
• Conflict detection across sources
• Conflict storage and resolution
• Reporting of unresolved conflicts
• Fully async backend

---

Author

Arun S Nair