Back-to-the-future motion analysis using machine intelligence predicts the potential risk of mood disorders

Abstract

Motion encodes emotional value. Decoding the hidden information within the complexity and randomness of motion may aid in the early diagnosis and prevention of mood disorders. In this study, we utilized machine intelligence to predict the risk of mood disorders by analyzing random motion patterns in mice. The support vector machine, trained on motion data collected before chronic social defeat stress (CSDS), predicted a significant correlation between these data and the degree of CSDS-induced social avoidance. Unsupervised clustering revealed key motion biomarkers of stress susceptibility, accounting for approximately 70% of the onset of social avoidance in a new group of mice before exposure to CSDS. These findings suggest that motion analysis using machine intelligence could provide a non-invasive approach for predicting the risk of mood disorders.