Combination of classification and forecasting methods to predict the possible occurrence of depressive episodes based on motor activity data
Abstract
While depressive disorder (or depression) is a common and treatable mental disorder, chronic depression causes great suffering to the affected person, causes deterioration in personal relationships, a reduction in the performance of treatments necessary to control the progression of a chronic illness or lead to suicide in the worst case. People's interest in monitoring their health has driven the development of devices to monitor blood pressure and heart rate, diet, and activity trackers, among others. This recent trend in the production of wearable devices, together with the growing development of Artificial Intelligence, has contributed to the detection and treatment of many diseases. Motor activity has been used to detect mental illness, however, research in the field has been limited to diagnosing the presence of a depressive disorder. This article presents a combination of classification and forecasting methods to predict the occurrence of depressive episodes. Model outcomes can allow the adoption of preventive treatments through emotional support actions to prevent mild depression from becoming a chronic disorder.