Exploration of a Machine Learning Model Using Self-rating Questionnaires for Detecting Depression in Patients with Breast Cancer

Abstract

Objective: Given the long-term and severe distress experienced during breast cancer treatment, detecting depression among breast cancer patients is clinically crucial. This study aimed to explore a machine-learning model using self-report questionnaires to screen for depression in patients with breast cancer. Methods: A total of 327 patients who visited the breast cancer clinic were included in this study. Depressive symptoms were measured using the Patient Health Questionnaire-9 (PHQ-9), Beck Depression Inventory (BDI), and Hospital Anxiety and Depression Scale (HADS). The depression was evaluated according to the Diagnostic and Statistical Manual of Mental Disorders 5th edition. The prediction model's performance based on supervised machine learning was con Results: The BDI showed an area under the curve (AUC) of 0.785 when using the logistic regression (LR) classifier. The HADS and PHQ-9 showed an AUC of 0.784 and 0.756 when using the linear discriminant analysis, respectively. The combinations of BDI and HADS showed an AUC of 0.812 when using the LR. The combinations of PHQ-9, BDI, Conclusion: The combination model with BDI and HADS in breast cancer patients might be better than the method using a single scale. In future studies, it is necessary to explore strategies that can improve the performance of the model by integrating the method using questionnaires and other methods.