Predictive Analysis Model to Improve the Control of Eating Disorders in Adolescents in Metropolitan Lima Based on Machine Learning

Objective Eating disorders in adolescents have been a recurring issue within our society, significantly impacting their mental and physical well-being. This study aims to implement a predictive analysis model based on machine learning and the EDI-3 evaluation method to forecast eating disorders (ED) in adolescents. Method Using the machine learning tool with the Random Forest algorithm and the Eating Disorder Inventory-3 (EDI-3) assessment method, a dataset of information was gathered from a sample of 500 adolescent students. This dataset will contribute to the training of the predictive model, enabling it to identify patterns of eating behaviors in end-users. Results Through data cleaning of the final dataset, 70% of the information was utilized for training the predictive model, and 30% was allocated for subsequent validation. This approach yielded an effectiveness rate of 92.27% upon completion of the training. Furthermore, in order to validate these results, Cronbach’s alpha coefficient was employed, resulting in a score of 0.70, indicative of a satisfactory level of reliability. Discussion The results obtained strongly support one of the main objectives of the study conducted, as it significantly surpasses the 85% accuracy threshold. Our results suggest that eating behavior patterns can be crucial factors in making predictions that enable the early identification of positive cases.