Mental Disorder Screening Via Wearable Internet Of Things: Deep Neural Network Approach

Mental health problems are becoming a significant international health issue, and a late diagnosis has, in most cases, caused serious social, economic, and clinical impacts. Early detection is consequently a requirement for timely intervention and better results. The latest developments in wearable Internet of Things (Iota) technology have enabled continuous, non-invasive measurements of physiological indicators, including heart rate variability, sleep behavior, and exercise intensity, and have opened new opportunities for objective mental health measurement. Nevertheless, traditional deep learning-based approaches to the detection of mental disorders are computationally expensive and cannot be deployed on wearable or edge computing devices with limited resources, as they have high latency and consume a lot of energy. This paper suggests deep neural network (DNN) architecture for screening mental disorders using wearable Iota data. The approach to it is pre-processing a set of multi-sensor physiological data, obtaining compact feature representations, and training an optimized DNN with fewer parameters. Accuracy, precision, recall, F1-score, model size, and inference time are the measures of model performance. The suggested solution exhibits accuracy in competitive screening but minimizes the computational load, which is why it is appropriate in the digital application of mental health wearables in real-time.