Detection and Diagnosis of Different Types of Obesity Using Machine Learning-Based Approaches

Abstract

With the escalating global prevalence and associated health risks of obesity, there is a critical need to detect the condition at its earliest stages and anticipate its diverse forms. Traditional methods may be insufficient, necessitating the exploration of advanced approaches, such as machine learning, to analyze extensive health data and enhance our ability to forecast obesity outcomes accurately. The primary objective of this study is to leverage machine learning techniques for the early detection and accurate prediction of various forms of obesity. By utilizing a machine learning-based method, the aim is to estimate the risk of obesity and contribute to the development of effective preventative and therapeutic measures. This study contributes valuable insights to the field by demonstrating the efficacy of machine learning methodologies in forecasting different forms of obesity. By utilizing a benchmark dataset and employing various machine learning methods, including KNN, random forests, naïve bayes, support vector machines, and logistic regression, the research provides a comprehensive evaluation of model performance. from Kaggle includes seventeen parameters related to stress. Assessment measures include precision, loss, recall, accuracy, and F1-score. The random forest classification algorithm exhibits superior accuracy rates of 97% for Obesity_Type_I, 99% for Obesity_Type_II, and 83% for Obesity_Type_III compared to alternative classifiers. The findings contribute to existing knowledge on the application of machine learning for predicting obesity, offering potential improvements in strategies for prevention and management.

Topics & Concepts