Predicting Executive Function Impairments in Young Adults Using Machine Learning and Lifestyle Data

The development of executive function (EF) impairments in young individuals, such as difficulties with attention, memory, and problem-solving, is influenced by biological, social, and lifestyle factors. However, research on predicting these impairments remains limited due to a lack of reliable tools. This study analyzed 90 university students using EF tests, lifestyle, and sociodemographic questionnaires. Five machine learning models were evaluated: Decision Trees (DT), k-Nearest Neighbors (KNN), Support Vector Machines (SVM), Logistic Regression (LR), and Random Forest (RF), with cross-validation applied for model assessment. The results indicated a 62% incidence of EF impairments. Maternal education and nutrition were identified as key influencing factors. Among the models, DT performed best, achieving a recall of 61.9%, an F1-score of 62.1%, and an AUC of 66.54%, while RF had the lowest performance. Limitations include the cross-sectional nature of the data, which restricts causal inference, and the reliance on self-reported responses from participants, which may reduce data reliability. Despite these limitations, this study demonstrates the feasibility of using machine learning to predict EF impairments based on easily collected sociodemographic and lifestyle data. Sociodemographic and lifestyle variables are valuable predictors of EF impairments in young individuals. Machine learning tools offer a practical approach to assessing population-level EF health using accessible data.