Undisturbed Mental State Assessment in the 5G Era: A Case Study of Depression Detection Based on Facial Expressions

Abstract

5G technology brings a comprehensive improvement in the network layer, which meets real-time, high-efficiency, and stability requirements in medical scenarios to a large extent, such as remote diagnosis and surgery. The heavy burden and severe impact of mental disorders make it desirable to find quantitative and automatic assessment approaches for early-stage detection of mental disorders. Facial expressions contain abundant emotional information, which may reflect abnormal mental states like anxiety and depression. With low latency and high bandwidth, 5G makes real-time monitoring of mental health feasible. In this article, a novel undisturbed mental state assessment prototype is proposed, which uses facial video streaming collected with 5G terminals to assess the mental state of a user in real time. A case study of depression detection using facial expressions has been developed based on the prototype. As a study case, we collected facial expression data from patients with depression and healthy people as control subjects. We extracted the transitional optical flow under stimulus feature and used the decision tree for classification. Results show that our depression assessment model is effective, and also reflect the feasibility and validity of our prototype.