Monitoring Motor Activity Data for Detecting Patients' Depression Using Data Augmentation and Privacy-Preserving...

Area of Science:

• Digital Health
• Computational Psychiatry
• Biomedical Signal Processing

Background:

• Wearable devices collect personalized physiological data for healthcare applications.
• Motor activity signals from devices like ActiGraph wristbands show potential for depression detection.
• Accurate depression classification models require substantial, sensitive, multi-subject data.

Purpose of the Study:

• To develop a privacy-preserving data augmentation technique for motor activity signals.
• To improve the accuracy of depression classification models using enhanced data.
• To evaluate the proposed method against existing state-of-the-art techniques.

Main Methods:

• Utilized motor activity data from ActiGraph wearable wristbands.
• Developed a novel data augmentation technique preserving data privacy.
• Extracted classification models for depression prediction.
• Evaluated performance on the Norwegian INTROMAT Project mental health dataset.

Main Results:

• The proposed augmentation technique enhances the volume and utility of motor activity data.
• The developed classification models show improved performance in depression detection.
• The approach demonstrates effectiveness compared to current state-of-the-art methods.

Conclusions:

• Privacy-preserving data augmentation is a viable strategy for improving wearable-based mental health monitoring.
• This method facilitates the development of more accurate depression detection systems.
• The findings support the use of adaptive technology in mental health care.