A study on the relationship between orthorexia and lateral thinking using machine learning

Abstract

Introduction and objectives: Orthorexia, one of the most important health problems today, is an unhealthy attachment to a healthy diet. This study aimed to investigate the relationship between lateral thinking disposition and orthorexia using classical statistical methods and machine learning. Material and methods: This descriptive study was conducted between June and October 2020 with nursing students at a college in eastern Turkey. Relationship between lateral thinking and orthorexia and agreement between classical statistical methods and machine learning methods. Results: BMI, gender, predisposition to lateral thinking, and family income were identified as important predictors of orthorexia. The deep learning algorithm provided more efficient results than the regression analysis and other machine learning algorithms (accuracy=66.0%, AUC=0.71). Conclusions: In eating disorder studies, such as ON, machine learning can help interpret complex relationships more accurately.