Effect of Obesity in Disease Transmission Dynamics using Machine Learning

Overview

1. Individual Risk Prediction: Predicting stroke and heart disease from clinical features including BMI/obesity status
2. Population Transmission Dynamics: Simulating and forecasting epidemic spread with obesity as a comorbidity risk modifier

Datasets

• Stroke Prediction Dataset (Nnaodeh/Stroke_Prediction_Dataset)
  • 5,110 patient records with 11 clinical features
  • BMI used as obesity proxy (BMI ≥ 30 = obese)
  • Target variables: stroke and heart_disease

Methods

Individual Disease Risk Models

• Logistic Regression (best for stroke: AUC = 0.839)
• Random Forest
• Gradient Boosting
• XGBoost
• Multi-Layer Perceptron (Neural Network)

Epidemic Simulation

• Agent-based SIR model (Susceptible-Infected-Recovered)
• Population: 10,000 individuals
• Obesity susceptibility multiplier: 2.0x
• Base infection rate: 0.03 per contact
• Recovery rate: 0.08 (~12 day infectious period)

Epidemic Forecasting Models

• Ridge Regression
• Random Forest Regressor
• Gradient Boosting Regressor
• XGBoost Regressor (best: MAE = 3.78 for total infections)

Key Findings

Obesity Effect on Individual Disease Risk

Obesity Effect on Epidemic Transmission

• Obese attack rate: 96.49%
• Non-obese attack rate: 78.97%
• Relative risk: 1.22x
• Peak infected day: Day 44 (3,816 simultaneous infections)

Feature Importance (Stroke Prediction)

1. Age (most important)
2. Average glucose level
3. BMI / Obesity status

Files

• obesity_disease_transmission.py — Complete pipeline script
• report.json — Detailed metrics and results
• results/*.png — Visualizations

Reproduction

Citation

• Stroke Prediction Dataset: Nnaodeh/Stroke_Prediction_Dataset
• Christakis & Fowler (2007): "The Spread of Obesity in a Large Social Network over 32 Years"

Generated by ML Intern

• Try ML Intern: https://smolagents-ml-intern.hf.space
• Source code: https://github.com/huggingface/ml-intern

Usage