Multi-modal data collection for measuring health, behavior, and living environment of large-scale participant cohorts

Abstract

Background As mobile technologies become ever more sensor-rich, portable, and ubiquitous, data captured by smart devices are lending rich insights into users’ daily lives with unprecedented comprehensiveness and ecological validity. A number of human-subject studies have been conducted to examine the use of mobile sensing to uncover individual behavioral patterns and health outcomes, yet minimal attention has been placed on measuring living environments together with other human-centered sensing data. Moreover, the participant sample size in most existing studies falls well below a few hundred, leaving questions open about the reliability of findings on the relations between mobile sensing signals and human outcomes. Results To address these limitations, we developed a home environment sensor kit for continuous indoor air quality tracking and deployed it in conjunction with smartphones, Fitbits, and ecological momentary assessments in a cohort study of up to 1,584 college student participants per data type for 3 weeks. We propose a conceptual framework that systematically organizes human-centric data modalities by their temporal coverage and spatial freedom. Then we report our study procedure, technologies and methods deployed, and descriptive statistics of the collected data that reflect the participants’ mood, sleep, behavior, and living environment. Conclusions We were able to collect from a large participant cohort satisfactorily complete multi-modal sensing and survey data in terms of both data continuity and participant adherence. Our novel data and conceptual development provide important guidance for data collection and hypothesis generation in future human-centered sensing studies.

Gaussian Latent Dirichlet Allocation for Discrete Human State Discovery

“…Ground truth for such momentary fluctuation is not available in the datasets used in this article but we defer to future work to validate the utility of these momentary states for predictive, real-time monitoring. Increasingly available, dense, and multi-modal passive sensing data from mobile, wearable, and environmental sensing devices [11] should provide plenty of evidence for in-the-moment objective ground truth of an individual's daily behavior and health.…”

circadian rhythms are not captured equal: Exploring Circadian metrics extracted by differentcomputational methods from smartphone accelerometer and GPS sensors in daily life tracking

“…The data we use for this study come from the UT1000 Project 18 we conducted at the University of Texas at Austin in two deployments, one in the Fall 2018 semester (October-November) and the other during Spring 2019 semester (February-March). This project was approved by the University of Texas Institutional Review Board (study number 2018-07-0035; approval date: September 25, 2018; expiration: July 16, 2019).…”

“…We aim to address these limitations in this paper. We base our analyses on smartphone tracking data we collected from 1,584 college students (225 participants’ data were retained in the analyses after preprocessing) in a major public university in the United States over three weeks in the 2018-2019 school year 18 . We conduct CR analysis using the four main types of methods outlined above (survey construct automation, cosinor, non-parametric, Fourier) on two types of smartphone sensor signals (accelerometer, GPS), and compare the same metrics extracted from different sensor signals as well as compare different metrics extracted from the same sensor signal.…”

Mobile Sensing Methods

“…However, as the use of other technical devices is bound to increase in mobile sensing research (see, for example, the UT1000 project combining mobile sensing via smartphones and fitness trackers by Wu et al, 2021), we include in this chapter discussions of such devices where good psychological research examples are available (Mehl et al, in press).…”