A machine learning approach to optimize the assessment of depressive symptomatology

In this work, we developed a machine learning model to predict depressive symptomatology (DS). The model achieved an AUC of 0.71 and a sensitivity of 74.8%. When developing the model, we applied a Bayesian network approach to select its predictors. This probabilistic approach facilitates the understanding of the relationships between predictors and DS. In consequence, we were able to identify that having balance problems and experiencing shortness of breath are directly related to DS.