Quantifying Lifestyle and Behavioral Predictors of Obesity Risk: A Multinomial Logistic Research Approach

Obesity is a major risk factor for various health conditions, including cardiovascular disease, diabetes, respiratory problems, and certain cancers. In Latin America, obesity rates continue to rise, contributing to increased morbidity and mortality. While previous research has utilized machine learning models to classify obesity levels, these models do not quantify the impact of lifestyle and behavioral factors on obesity risk. This study aimed to quantify the significance of individual predictors using multinomial logistic regression models. We analyzed the Estimation of Obesity Levels Based On Eating Habits and Physical Condition dataset from the UCI Machine Learning Repository, which included 2,111 Latin American individuals. The dataset covers seven obesity levels and factors such as dietary habits, physical condition, and demographics. After refining the models to include statistically significant predictor (p<0.05), we found seven key factors, including frequent consumption of high-calorie food (FAVC), meal frequency (CAEC), self-monitoring of calorie intake (SCC), technology usage (TUE), standardized age (Z score Age), and standardized BMI (Z-score of BMI). All of these factors were statistically significant (p<0.001). Standardized BMI showed the strongest association with obesity risk across all categories except for insufficient weight. These findings highlight the importance of lifestyle and behavioral factors in obesity risk. Although our study is limited to three Latin American countries, the methodology can be applied to other populations. This approach offers a framework for future research on obesity risk and the development of targeted interventions in diverse settings.