Functional Data Analysis Approaches for Analyzing Mobile Health Data

• Activity Number: 567 - Digital Phenotyping – What Can Wearables and Smartphones Tell Us About Our Mental Health?
• Type: Topic Contributed
• Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Jihui Lee* and Samprit Banerjee
• Companies: Weill Cornell Medicine and Weill Medical College, Cornell University
• Keywords: Mobile health data; Ecological momentary assessment; Functional data analysis; Digital phenotyping

Abstract

Using mobile or wearable devices, such as smartphones and activity trackers, provide interactive interface for ecological momentary assessment (EMA) as well as collect vast amount of behavioral measures using passive sensing in a more natural environment than laboratory. The EMA linked with behavioral measures broadens the understanding of gradual progress and treatment effects in depression studies. We aim to understand the association between the self-reported level of depression and behavioral trajectories of elderly patients with depression using smartphones. We apply functional data analysis (FDA) approaches to visualize and analyze the longitudinally recorded data; we conduct functional principal component analysis to identify the common structure of behavioral trajectories across patients and fit functional regression models to investigate the relationship between EMAs (self-reports) and behavioral measures. Results of FDA can be directly used to augment the assessment of patients’ behaviors and provide an interactive intervention tailored to each individual.