Early Detection of Respiratory Diseases with Random Forest

This project implements a Random Forest Regressor for the early detection of acute respiratory diseases (IRA) in Peru. The model predicts the weekly incidence rate by region, using climate and historical case data, while excluding the pandemic years (2020–2021).

⚙️ Data used

• Source 1: Observatorio de Clima y Salud – MINSA
• Source 2: Population from 2017 Census – INEI
• Time coverage: 2017–2025 (updated weekly)
• Geographic coverage: 26 regions of Peru (including Callao and Lima Metropolitana)
• Variables:
  • Climate: tmean, tmax, tmin, humr, ptot
  • IRA cases: Cases
  • Population by region (population)
  • Derived variables: incidence per 100k inhabitants, predicted rates, alert levels

🧠 Model

• Algorithm: RandomForestRegressor .

🔍 Features

• Weekly incidence predictions (Predicted_Rate)
• 🚨 Alert levels:
  • 🟢 Green → Low incidence
  • 🟡 Yellow → Moderate risk
  • 🔴 Red → High incidence
• Files ready for Power BI visualization (CSV/Excel)
• Graphs of variable importance and model performance

🛠️ Next Steps

• Integrate predictions into a live dashboard (Power BI).
• Extend forecasting to future weeks/years.

🔦 Collaborators

• Monica Lozada
• Carolina Saavedra