Detecting major depressive disorder presence using passively-collected wearable movement data in a...

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomedical Engineering

Background:

• Major Depressive Disorder (MDD) presents diagnostic challenges due to its heterogeneity.
• Sleep and movement pattern changes are potential early indicators for MDD detection.
• The Patient Health Questionnaire-9 (PHQ-9) is a standard screening tool for MDD.

Purpose of the Study:

• To investigate the effectiveness of wrist-worn actigraphy data for detecting MDD.
• To apply machine learning (ML) and deep learning (DL) techniques to actigraphy data for MDD identification.
• To identify specific actigraphy-derived biomarkers associated with MDD.

Main Methods:

• Utilized minute-level actigraphy data from 8,378 participants in the NHANES study.
• Employed two ML approaches: traditional ML with feature derivation and a deep learning Convolutional Neural Network (CNN) with Gramian Angular Field transformation.
• Compared the efficacy of these ML strategies in detecting MDD.

Main Results:

• Movement-related features were most influential in the traditional ML approach for MDD detection.
• Nighttime movement emerged as the most significant feature in the CNN approach for identifying MDD.
• Actigraphy data, analyzed via ML, demonstrated potential for MDD detection.

Conclusions:

• Passively collected actigraphy data holds significant potential for improving the understanding, diagnosis, and treatment of MDD.
• ML and DL techniques can effectively leverage actigraphy data to identify individuals with MDD.
• Movement patterns captured by actigraphy serve as valuable biomarkers for MDD.