Automatic prediction of depression in older age

Maintaining good mental health such as the prevention of severe depressive symptoms is critical for physical health and well-being in older adulthood. However, depression in elderlies is not known quite well and thus cannot be treated adequately. In this study, a large and wide variety of influencing factors from multiple domain areas were investigated using a large nationally representative sample of older people from the English Longitudinal Study of Ageing (ELSA). Five different machine learning algorithms were employed to build the models for the prediction of depression in older age. Several model ensemble strategies were proposed to merge the results from individual predictive models in order to further improve prediction performance. Significant risk or protective factors associated with depressive symptoms in the elder were separately identified in each domain area. The findings from this study will enhance our understanding about the underlying pathophysiology of depression, thus helping develop appropriate intervention strategies to prevent or reduce the onset of depression in older age.