Clinical Copilot is a state-of-the-art medical decision support system designed for high-density clinical environments. It leverages multiple specialized ML models for risk prediction and a Large Language Model (Gemini 1.5 Flash) for clinical reasoning and diagnosis synthesis.
- X-Terminal Dashboard: A high-density, Bloomberg-style UI designed for rapid data interpretation.
- Multi-Model Risk Engine:
- Heart Disease Risk: XGBoost model trained on clinical cardiovascular data.
- Diabetes Risk: Classifier based on long-term biometric trends.
- Stroke Assessment: Predictive analysis of vascular health.
- Kidney Health AI: Specialized indicators for chronic kidney disease risk.
- Neural Differential Assessment: LLM-powered (Gemini) synthesis that explains ML findings in clinical language, identifies paradoxes, and suggests next steps.
- RAG-Lite Semantic Search: Injects past doctor feedback and similar historical cases into the LLM prompt for context-aware reasoning.
- Medication Safety HUD: Real-time drug-drug and drug-condition interaction scanning with dynamic risk categorization.
- Clinical Confidence Score: Statistical confidence metrics for every prediction.
- Live System Telemetry: Real-time monitoring of network latency and microservice performance.
- Biometric Telemetry: Live streaming of BP, Glucose, BMI, Cholesterol, Heart Rate, and Steps.
- New Triage System: Seamless entry of new patient data with instant analysis.
- 🛡️ Blockchain Audit Layer: High-performance, internal ledger that hashes every decision for regulatory compliance (FDA PCCP).
Tip
New in v2.0: See System Upgrades for a deep dive into the latest architectural and feature enhancements.
The system follows a high-performance distributed architecture:
- Frontend (Next.js 14 + Tailwind + Framer Motion):
- Highly responsive viewport with glassmorphism aesthetics.
- Blockchain Verification UI for real-time audit proofs.
- Backend Orchestrator (Go Fiber):
- Handles patient persistence (PostgreSQL + GORM).
- Internal Blockchain Engine: Custom SHA-256 implementation for immutable logging.
- Manages RAG-lite context retrieval.
- Coordinates requests between Frontend and ML microservices.
- ML Microservice (Python FastAPI):
- Serves multiple
joblibserialized scikit-learn/XGBoost models. - Handles feature transformation and alignment for inconsistent datasets.
- Integrates Google Generative AI (Gemini) for clinical reasoning.
- Serves multiple
- Languages: Go, Python, TypeScript (TSX)
- Frameworks: Fiber (Go), FastAPI (Python), Next.js (Web)
- ML/DS: Scikit-Learn, XGBoost, Pandas, Numpy
- Database: SQLite (Relational + Local Persistence)
- LLM: Google Gemini 1.5 Flash (Clinical Synthesis)
- Styling: Tailwind CSS, Lucide Icons, Framer Motion (Animations)
- Go 1.21+
- Python 3.10+
- Node.js 18+
- Gemini API Key (stored in
.env)
Create a .env file in the root:
GEMINI_API_KEY=your_key_herecd src/api/ml_api
pip install -r requirements.txt
python main.pycd backend
go run cmd/server/main.gocd frontend
npm install
npm run dev -- -p 4000Access the terminal at http://localhost:4000.
Clinical Copilot is a clinical decision support tool (CDST) and should only be used as an auxiliary aid. Final diagnoses and decisions must be made by qualified medical professionals.