Machine-Learning Predictive Models For Dependency On Smartphones Based On Risk Factors

Abstract

Abstract Background: Machine learning techniques allow highly accurate prediction of different tasks by measuring the event probabilities. This research proposes a prediction model for dependency on smartphones based on machine learning techniques. Methods: We performed an analytical observational study with a retrospective case–control approach; the different classification methods used were decision tree, random forest, logistic regression, and support vector machine. The sample demographic included 1228 students from a private university in Cali. The tests were 1) smartphone dependency assessment and 2) the Nordic musculoskeletal symptoms questionnaire. Results: It was found that some of the variables related to smartphone dependency are academic curriculum, school, marital status, socioeconomic status, rules, discussions, and discrimination. Conclusions: The support vector machine model evidences highest prediction precision for smartphone dependency, obtained through the stratified-k-fold cross-validation technique.