Detection of Depression from IoMT Time Series Data using UMAP features

Depression is the most serious and prevalent public health problem that has led to an increase in suicidal risks, suicidal ideations, and much more impact on family relations. A diagnosis of mental health problems lacks a direct quantitative test. Conventionally experts evaluate the symptoms and signs through patient interviews or questionnaires. Data generated by wearable IoMT devices (e.g. wrist actigraph) provides an efficient objective way to diagnose depression and related mental health problems in the early stages. In this paper, we propose UMAP (Uniform Manifold Approximation and Projection) based feature extraction and neural network classifier for depression detection (both Unipolar and bipolar disorders) from IoMT (Internet of Medical Things) data (Depresjon Dataset). The classification of depression from 24-hour motor activity data shows a sensitivity of 96.257% and a specificity of 98.50%.