Real-Time Medical Risk Assessment & Adaptive Questionnaire System
MediRisk AI is an end-to-end system that processes medical responses in real-time, generates risk scores, adapts questionnaires dynamically,Mental health check up and creates actionable insurance reports. Designed for insurers and clinicians, it integrates medical guidelines and provides underwriters with intuitive risk insights.
Key Innovations:
- Real-time risk scoring with explainable AI.
- risk assesment report for the insurance company based on the user's data
- Reinforcement Learning (RL)-driven adaptive questioning.
- Mental Health check up and review.
- HIPAA-compliant underwriter dashboard.
- Dynamic Questioning: Adapts follow-up questions using RL and decision trees.
- Mental Health Score: User is asked Pyscometric Questions and then scores and recommendations are assigned on his mental health.
- Live Risk Meter: Visualizes risk score evolution during the chat.
- Data Extraction from Report: extract the body report which is uploaded by user and helps to make questions on it.
- Models: XGBoost (structured data), BioBERT (NLP), Survival Analysis (long-term risk).
- Guideline Compliance: Flags deviations from CDC/WHO/ACC-AHA rules.
- Risk Heatmaps: Geographic distribution of high-risk cases.
- Audit Trails: Track AI decisions vs. human overrides.
- Automated Reports: One-click PDF/Excel generation.
-
Backend:
- Real-Time API: FastAPI,Node.js, express.js .
- ML Models: XGBoost for risk prediction, Hugging Face for NLP.
- Database: PostgreSQL (structured), Elasticsearch (text).
-
Frontend:
- Chat Interface: React.js css .
- Dashboard: Plotly/Dash for visualizations.
-
Deployment:
- Cloud: AWS/Azure (HIPAA-compliant buckets).
- DevOps: Kubernetes, GitHub Actions CI/CD.
- Python 3.9+
- PyTorch, Transformers, XGBoost
- React.js, D3.js, Plotly
- PostgreSQL, Elasticsearch