Multi-Disease Risk Prediction Project

This project predicts patient risk for three diseases using machine learning:

• Diabetes
• Heart disease
• Kidney disease

It includes dataset download, model training, saved model files, and final performance metrics.

Objective

Build a screening-oriented ML system that identifies whether a patient is at risk for major diseases from clinical attributes.
This is a decision-support prototype and not a medical diagnosis system.

What Has Been Implemented

• diabetes_risk_project.ipynb: detailed diabetes workflow (EDA, preprocessing, training, evaluation, explainability)
• multi_disease_risk_project.ipynb: multi-disease analysis and comparison
• download_train_export.py: downloads datasets, trains models, saves artifacts, exports performance index
• diabetes_risk_report.md: deep technical report
• project_objective_and_work_done.pdf: concise project summary PDF

Project Structure

• datasets/ -> downloaded CSV files
• models/ -> trained .joblib models
• performance/ -> exported performance metrics and model index
• results_package/ -> reusable package for reading and summarizing outcomes

Datasets Used

• Diabetes: datasets/diabetes.csv
• Heart: datasets/heart.csv
• Kidney: datasets/kidney.csv

Source URLs are defined in download_train_export.py.

Training and Evaluation Procedure

1. Load disease dataset
2. Clean and preprocess features (imputation + scaling)
3. Train candidate models (Logistic Regression, Random Forest)
4. Select best model with cross-validation ROC-AUC
5. Calibrate probabilities using CalibratedClassifierCV
6. Evaluate on holdout test split
7. Save model and export metrics

Outcome Metrics

The following metrics are generated in performance/performance_index.csv:

• ROC-AUC
• PR-AUC
• F1-score
• Precision
• Recall
• Confusion values (tp, tn, fp, fn)

Model Performance (Current Run)

Disease	Model	ROC-AUC	PR-AUC	F1	Precision	Recall
Diabetes	Logistic Regression	0.8117	0.6716	0.5361	0.6047	0.4815
Heart	Random Forest	0.9026	0.9202	0.8649	0.7805	0.9697
Kidney	Logistic Regression	1.0000	1.0000	1.0000	1.0000	1.0000

How to Run

Install dependencies:

pip install -r requirements.txt

Train models and generate outputs:

python download_train_export.py

Generate a compact summary report from metrics package:

python -m results_package.metrics_report

Output Files After Run

• models/diabetes_model.joblib
• models/heart_model.joblib
• models/kidney_model.joblib
• performance/performance_index.csv
• performance/performance_index.json
• performance/model_index.json
• performance/results_summary.md (from package)

Important Note

This project is for risk screening and educational/research purposes.
Any clinical decision must be reviewed by qualified medical professionals.