Related Experiment Video
Updated: Sep 4, 2025
07:25
An Objective and Child-friendly Assessment of Arm Function by Using a 3-D Sensor
Published on: February 12, 2018
7.0K
Depression and Severity Detection Based on Body Kinematic Features: Using Kinect Recorded Skeleton Data of Simple
Yanhong Yu1, Wentao Li2, Yue Zhao3
1College of Traditional Chinese Medicine, Shandong University of Traditional Chinese Medicine, Jinan, China.
Frontiers in Neurology
|July 18, 2022
Summary
This study introduces a novel method using Kinect V2 skeletal data and a spatial attention dilated temporal convolution network (SATCN) to detect depression. The system achieved up to 75.8% accuracy in classifying depression severity, aiding patient screening and recovery monitoring.
Area of Science:
- * Computational psychiatry and affective computing.
- * Biomedical engineering and human-computer interaction.
Background:
- * Assessing depression often relies on subjective measures, highlighting the need for objective biomarkers.
- * Relative limb movement patterns are recognized as potential indicators of depressive states.
- * Previous methods for analyzing movement data in depression assessment have limitations.
Purpose of the Study:
- * To investigate the efficacy of a skeleton-mimetic task with natural stimuli for depression recognition.
- * To develop and validate a novel deep learning model for automated depression detection using skeletal data.
- * To assess the model's capability in identifying depression severity and monitoring patient recovery.
Main Methods:
- * Utilized Kinect V2 for collecting sequential 3D skeletal data from 25 body joints.
- * Developed a Spatial Attention Dilated Temporal Convolution Network (SATCN) integrating temporal convolution groups and spatial attention.
- * Trained and evaluated the SATCN model on binary (depression vs. non-depression) and multi-class (depression severity) skeletal datasets.
Main Results:
- * The SATCN model achieved a maximum accuracy of 75.8% for binary classification of depression.
- * Multi-class classification for depression severity reached an accuracy of 64.3%.
- * The model demonstrated potential for fine-grained identification of depression levels and tracking recovery progress.
Conclusions:
- * Kinect V2-based skeletal data analysis combined with the SATCN model offers a promising objective approach for depression screening.
- * The method can assist clinicians in identifying patients and monitoring their recovery trajectory.
- * This technology has the potential to improve the accessibility and objectivity of mental health assessments.
