Diabetes Risk
Prediction System

Powered by a novel Hybrid RF-GBDT model and 7 state-of-the-art machine learning algorithms for highly accurate diabetes risk assessment.

DIABETES_PREDICTION_PIPELINE
dataset PIMA_768_records
models_trained 7 / 7 complete
best_accuracy 75.76%
best_auc_roc 0.8394
pipeline_status READY
hybrid_rf_gbdt DEPLOYED
768
Patients Analyzed
7
ML Models Compared
83.9%
Best AUC-ROC Score

What You Can Do

Explore powerful tools backed by cutting-edge research

Predict Risk

Enter 8 simple health parameters and instantly receive a diabetes risk prediction from multiple ML models.

Try Now
Compare Models

Explore an interactive dashboard comparing 7 ML models on accuracy, precision, recall, F1 score, and ROC-AUC.

View Dashboard
Research-Backed

Built on a peer-reviewed research paper proposing a novel Hybrid RF-GBDT ensemble model for diabetes prediction.

Learn More

About the Project

This system is developed as part of a B.Tech final-year research project at Raj Kumar Goel Institute of Technology. It implements a comprehensive machine-learning pipeline for diabetes risk prediction using the internationally recognized PIMA Indians Diabetes Dataset.

Seven classification algorithms are trained and compared, including traditional models, popular ensembles, and a novel Hybrid RF-GBDT model that achieves state-of-the-art accuracy.

Read the Research

Hybrid RF-GBDT Ensemble Architecture

Ready to Assess Your Diabetes Risk?

Enter your health parameters and get instant predictions from 7 ML models