Using an improved relative error support vector machine for body fat prediction.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Health Informatics

Background:

• Obesity, characterized by excessive body fat, is a chronic disease with significant health complications.
• Current body fat estimation methods are often costly and require specialized equipment.
• Accurate, accessible body fat percentage prediction is vital for obesity assessment and management.

Purpose of the Study:

• To develop a cost-effective approach for predicting body fat percentage using easily accessible body measurements.
• To improve the accuracy of body fat prediction models.

Main Methods:

• An improved relative error support vector machine (SVM) approach was developed.
• A bias error control term was incorporated into the SVM objective function for unbiased estimation.
• Feature selection was employed to enhance prediction accuracy by removing redundant or irrelevant features.
• The Wilcoxon rank-sum test was used for statistical validation.

Main Results:

• The proposed SVM method demonstrated superior performance compared to other prediction models across four evaluation metrics.
• Feature selection was shown to further enhance prediction performance.
• Statistical analysis confirmed the significant superiority of the proposed method.

Conclusions:

• A novel and effective approach for predicting body fat percentage has been presented.
• The method provides a valuable tool for individuals to estimate body fat using accessible measurements.
• The Wilcoxon rank-sum test validated the significant performance improvement and the utility of feature selection.