Alert button

Using Convolutional Variational Autoencoders to Predict Post-Trauma Health Outcomes from Actigraphy Data

Nov 14, 2020
Ayse S. Cakmak, Nina Thigpen, Garrett Honke, Erick Perez Alday, Ali Bahrami Rad, Rebecca Adaimi, Chia Jung Chang, Qiao Li, Pramod Gupta, Thomas Neylan, Samuel A. McLean, Gari D. Clifford

Figure 1 for Using Convolutional Variational Autoencoders to Predict Post-Trauma Health Outcomes from Actigraphy Data
Figure 2 for Using Convolutional Variational Autoencoders to Predict Post-Trauma Health Outcomes from Actigraphy Data
Figure 3 for Using Convolutional Variational Autoencoders to Predict Post-Trauma Health Outcomes from Actigraphy Data
Figure 4 for Using Convolutional Variational Autoencoders to Predict Post-Trauma Health Outcomes from Actigraphy Data

Share this with someone who'll enjoy it:

Depression and post-traumatic stress disorder (PTSD) are psychiatric conditions commonly associated with experiencing a traumatic event. Estimating mental health status through non-invasive techniques such as activity-based algorithms can help to identify successful early interventions. In this work, we used locomotor activity captured from 1113 individuals who wore a research grade smartwatch post-trauma. A convolutional variational autoencoder (VAE) architecture was used for unsupervised feature extraction from four weeks of actigraphy data. By using VAE latent variables and the participant's pre-trauma physical health status as features, a logistic regression classifier achieved an area under the receiver operating characteristic curve (AUC) of 0.64 to estimate mental health outcomes. The results indicate that the VAE model is a promising approach for actigraphy data analysis for mental health outcomes in long-term studies.

View paper onarxiv icon

Share this with someone who'll enjoy it: