2020/11/14 by Ayse S. Cakmak, Nina Thigpen, Cakmak, Ayse S. +23
Psychology · #Digital Mental Health Interventions #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Mental Health Research Topics #Mental Health via Writing #Signal Processing (eess.SP) #Sleep and related disorders #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2011.07406
openalex publication_date 2020/11/14 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Depression and post-traumatic stress disorder (PTSD) are psychiatric\nconditions commonly associated with experiencing a traumatic event. Estimating\nmental health status through non-invasive techniques such as activity-based\nalgorithms can help to identify successful early interventions. In this work,\nwe used locomotor activity captured from 1113 individuals who wore a research\ngrade smartwatch post-trauma. A convolutional variational autoencoder (VAE)\narchitecture was used for unsupervised feature extraction from four weeks of\nactigraphy data. By using VAE latent variables and the participant's pre-trauma\nphysical health status as features, a logistic regression classifier achieved\nan area under the receiver operating characteristic curve (AUC) of 0.64 to\nestimate mental health outcomes. The results indicate that the VAE model is a\npromising approach for actigraphy data analysis for mental health outcomes in\nlong-term studies.\n