Advanced supervised machine learning methods for precise diabetes mellitus prediction using feature selection.

Area of Science:

• Medical Informatics
• Computational Biology
• Machine Learning in Healthcare

Background:

• Diabetes mellitus (DM) is a widespread chronic metabolic disorder with severe health implications if untreated.
• Early diagnosis of diabetes is crucial for timely intervention and preventing complications like blindness and kidney failure.
• Machine learning techniques (MLT) offer powerful tools for pattern recognition and disease prediction.

Purpose of the Study:

• To conduct a comparative analysis of supervised MLT for early diabetes prediction.
• To evaluate the performance of Support Vector Machine (SVM), Naïve Bayes (NB), K-Nearest Neighbors (KNN), and Random Forest (RF) using the Pima Indian Diabetes dataset (PIDD).
• To highlight the effectiveness of different algorithms and the importance of cross-validation in medical prediction.

Main Methods:

• Utilized the Pima Indian Diabetes dataset (PIDD) from the UCI repository.
• Employed a 10-fold cross-validation approach to address class imbalance and ensure generalizability.
• Evaluated model performance using accuracy, precision, recall, and F1-score.

Main Results:

• Support Vector Machine (SVM) demonstrated the highest accuracy at 91.5%.
• Random Forest (RF) achieved 90% accuracy, followed by K-Nearest Neighbors (KNN) at 89%, and Naïve Bayes (NB) at 83%.
• The results indicate significant variation in model performance based on the chosen algorithm.

Conclusions:

• SVM is highly effective for early diabetes prediction.
• This study underscores the importance of robust validation methods like cross-validation in medical predictive modeling.
• The findings provide a framework for selecting optimal models for real-world diabetes risk assessment.