Depression Detection using Minirocket and Deep Learning

Depression is a debilitating mental health disorder with a significant impact on individuals and society. For treatment and care to be effective, early detection and intervention are essential. This paper presents a detailed investigation into the use of MiniRocket transform in conjunction with deep learning techniques for depression detection using actigraphy data. Actigraphy, a non-invasive method for monitoring human activity patterns, offers a rich source of information that can be leveraged for mental health assessment. Our approach involves preprocessing the raw actigraphy data and applying the MiniRocket transform to extract highdimensional features. These features are then fed into a deep learning model consisting of a Convolutional Neural Network (CNN) for spatial feature learning and a Long Short-Term Memory (LSTM) network for temporal sequence modelling. The model is trained using a large dataset of actigraphy data collected from individuals with and without depression. We conduct extensive experiments to evaluate the effectiveness of our approach, comparing it with baseline methods and state-of-the-art techniques. Our findings show that the suggested method performs better in terms of sensitivity, specificity, and accuracy. Additionally, we examine the acquired, sensitivity, and specificity. Furthermore, we analyze the learned features to gain insights into the underlying patterns associated with depression. In therapeutic settings, the suggested approach may greatly enhance the early identification and tracking of depression. It provides a scalable and automated approach that can complement existing diagnostic methods, enabling more timely and personalized interventions for individuals at risk of depression.