Hybrid Majority Voting: Prediction and Classification Model for Obesity

Area of Science:

• Medical Science
• Computer Science
• Data Science

Background:

• Obesity is a significant global health issue linked to numerous chronic diseases.
• Obesity results from a complex interplay of genetic, physiological, environmental, nutritional, and lifestyle factors.
• Current diagnostic methods like Body Mass Index (BMI) have limitations, particularly for individuals with high muscle mass.

Purpose of the Study:

• To develop and evaluate an advanced machine learning model for precise obesity prediction and classification.
• To compare the performance of individual machine learning algorithms against a novel hybrid approach.

Main Methods:

• A hybrid majority voting model was developed, integrating Gradient Boosting Classifier, Extreme Gradient Boosting, and Multilayer Perceptron.
• Seven distinct machine learning algorithms were tested on open datasets from the UCI machine learning repository.
• The accuracy of individual models was compared to establish a baseline for the hybrid approach.

Main Results:

• The proposed majority voting-based hybrid model demonstrated a high accuracy of 97.16% in predicting and classifying obesity.
• The hybrid model significantly outperformed the accuracy of individual machine learning algorithms used in the study.
• The developed hybrid model also surpassed the performance of other hybrid models previously reported.

Conclusions:

• The novel hybrid machine learning model offers a highly accurate solution for obesity prediction and classification.
• This approach addresses limitations of traditional methods like BMI by leveraging advanced computational techniques.
• The findings suggest a promising direction for improving obesity diagnostics and management through machine learning.