Back to Projects
Technical Deep Dive
EpiScope
AI-driven platform for monitoring and predicting disease outbreaks using machine learning and RAG.
The Challenge
"Predicting disease distribution and understanding the 'why' behind outbreaks is difficult for healthcare decision-makers without specialized tools."
Project Overview
EpiScope enables hospitals to predict disease risk based on patient data, explore historical trends through dynamic visualizations, and understand predictions via SHAP explainability.
Key Features
XGBoost Prediction
High-accuracy prediction for target diseases like Malaria and Diabetes.
Interactive Dashboards
Dynamic exploration of Ghanian health hotspots using Plotly/Dash.
Model Explainability (SHAP)
Visualizes exactly which features influenced a specific prediction.
The Workflow
1
Clinician enters patient demographics and symptoms
2
XGBoost calculates probability score
3
SHAP generates feature importance graphs
4
Gemini provides a plain-English summary of the case
Technical Stack
| Category | Technologies |
|---|---|
| Backend | Django (Python), PostgreSQL |
| AI/ML | XGBoost, Google Gemini API, SHAP |
| Cloud | Google Vertex AI, Docker |
System Architecture
Hybrid AI architecture combining deterministic ML (XGBoost) with Generative AI (Gemini) for both precision and interpretation.
Project Structure
├── epi_app/ (Django App) ├── models/ (Serialized ML models) ├── dashboards/ (Plotly graphs) └── templates/ (Frontend views)
Development Roadmap
XGBoost Implementationcompleted
SHAP Integrationcompleted
Ghana Map Visualizationcompleted
LSTM Forecastingplanned