Biobank for mHealth: Collecting High-Frequency Mobile Sensor Data for Long-Lasting Research Utility

Archived Content

The Office of Behavioral and Social Science Research (OBSSR) archives materials older than three years that are no longer updated. This content is available for historical purposes, and the information and links may have changed over time.

Date: November 16, 2016, 2:00–3:00 p.m. ET
Location: Virtual
Presenter: Santosh Kumar, Ph.D.
Santosh Kumar, Ph.D.

Santosh Kumar, Ph.D.

Santosh Kumar, Ph.D.
Director, MD2K
Professor and Lillian & Morrie Moss Chair of Excellence in Computer Science
The University of Memphis

Overview

For long-lasting research utility, biomedical studies often archive biospecimens in biobanks so that they can be reprocessed to take advantage of future improvements in assays and support biomedical discoveries not possible at the time of data collection. mHealth studies, on the other hand, usually collect digital biomarkers (e.g., activity counts) that are specific to the computational models adopted by respective vendors at the time of data collection. This approach prevents any future validation of these biomarkers and makes it impossible to recompute newer biomarkers. To obtain similar long-lasting research utility as biobanks, raw sensor data must be collected that can be reprocessed in future to validate prior biomarkers and to obtain newer biomarkers. Doing so is, however, challenging due to high frequency, large volume, rapid variability, and battery life limitations of sensors and smartphones.

The Mobile Sensor Data-to-Knowledge (MD2K) Center of Excellence has successfully developed open-source software (for both mobile phones and the cloud) that allows collection of high-frequency raw sensor data. The smartphone software called mCerebrum supports concurrent collection of streaming data from 8+ wearable sensors, including Microsoft Band, MotionSense, EasySense, AutoSense, phone sensors (e.g., GPS), Omron weight and blood pressure, and Oral-B smart toothbrush. It supports high-frequency raw sensor data collection (at 800+ Hz for 70+ million samples/day), curation, analytics, storage (~2 GB/day), and secure uploads to a cloud. mCerebrum continuously assesses data quality to quickly detect and fix any data quality issues to correct sensor detachment or sensor misplacements on the body. Data science research conducted by MD2K has resulted in 10 mHealth biomarkers: stress, smoking, craving, eating, lung congestion, heart motion, location, activity, driving, and drug use. Several of these biomarkers are computed in real-time on the phone to support biomarker-triggered Just-in-Time Adaptive Interventions (JITAI). This webinar will present MD2K research and the software to support mHealth research studies.

Biography

Santosh Kumar, professor of computer science at The University of Memphis, has received numerous NIH grants from the National Institute of Dental and Craniofacial Research; the National Institute of Arthritis and Musculoskeletal and Skin Diseases; the National Institute on Aging, National Heart, Lung, and Blood Institute; the Office of the Director; the National Institute on Drug Abuse; and the National Institute of Environmental Health Sciences. He is the lead principal investigator and director for the NIH Center of Excellence for MD2K. His research utilizes both wearable and mobile-based sensors to inform new approaches to research, interventions, and the delivery of healthcare. Kumar's work embodies the Office of Behavioral and Social Sciences Research's goals of leveraging mHealth for new forms of data, methods, measures, and approaches to enhance the delivery, replicability, and scalability of health interventions.