An ensemble learning approach for diabetes prediction using boosting techniques.

Area of Science:

• Medical Informatics
• Computational Biology
• Machine Learning in Healthcare

Background:

• Diabetes is a global health challenge requiring early detection for effective management.
• Machine learning offers promising tools for enhancing disease diagnosis and prognosis.

Purpose of the Study:

• To develop and evaluate machine learning models for accurate early prediction of diabetes.
• To identify the most effective boosting algorithm for diabetes prediction using the Pima diabetes dataset.

Main Methods:

• Experiments were conducted using five boosting algorithms on the Pima diabetes dataset from UCI.
• Data preprocessing included exploratory data analysis, upsampling, normalization, feature selection, and hyperparameter tuning.
• Model performance was assessed using statistical metrics, k-fold cross-validation, and ROC curves.

Main Results:

• Gradient boosting demonstrated the highest prediction accuracy at 92.85%.
• The model's performance was further validated using precision, recall, and f1-score.
• The developed model showed superior accuracy compared to existing studies.

Conclusions:

• The gradient boosting model shows significant potential for early diabetes prediction.
• The approach can be adapted for predicting other diseases with similar indicators.
• This research supports healthcare providers with advanced diagnostic tools.