The JustWalk JITAI Study: A System Identification Experiment to Understand Just-in-Time States of Physical Activity
Study Details
Study Description
Brief Summary
The goal of this system identification experiment is to estimate and validate dynamical computational models that can be used in a future a multi-timescale model-predictive controller. System identification is an experimental approach used in control systems engineering, which uses random and pseudo-random signal designs to experimentally manipulate independent variables, with the goal of producing dynamical models that can meaningfully predict individual responses to varying provision of support. A system identification is single subject/N-of-1 experimental design, whereby each person is their own control. This 9-month system identification experiment will experimentally vary daily suggested step goals and provision of notifications meant to inspire bouts of walking during different plausible just-in-time states. Results of this system identification experiment will then enable the development a future multi-timescale model-predictive controller-driven just-in-time adaptive intervention (JITAI) intended to increase steps/day. The system identification experiment will be conducted among N=50 inactive, adults aged 21 or over who have no preexisting conditions that preclude them from engaging in an exercise program, as determined using the physical activity readiness questionnaire.
| Condition or Disease | Intervention/Treatment | Phase |
|---|---|---|
|
N/A |
Detailed Description
N=50 English-speaking adults aged 21+ who are physically inactive (self-reported engagement in less than 60 minutes/week of moderate-intensity activity) and own a smartphone (iPhone or Android) will be recruited. Participants will be provided with and asked to wear a Fitbit Versa 3 and use the study app, JustWalk, for 270 days.
A system identification experiment, which is a single subject/N-of-1 experimental protocol used in control systems engineering, will be conducted. This study is designed to empirically optimize dynamical models that can be used within a future model-predictive controller-driven just-in-time adaptive intervention (JITAI). This system identification experiment will include two experimentally manipulated components: 1) notifications delivered up to 4 time per day designed to increase a person's steps within the next 3 hours via either increased awareness of the urge to walk or via bout planning; and 2) adaptive daily step goal suggestions. Both components will be experimentally manipulated using procedures appropriate for system identification. Specifically, notifications prompting planning of short walks within the next 3 hours will be experimentally provided or not across variations of need (i.e., whether daily step goals were previously met), opportunity (i.e., the next three hours is a time window when a person previously walked), and receptivity (i.e., person received fewer than 6 messages in the last 72 hours and walked after notifications were sent). This enables experimental manipulation of varying "just-in-time" states, thus providing valuable data for guiding future predictions about when, where, and for whom a bout notification would produce the desired effects compared to not. Thus, this is a hypothesis-driven approach to better understanding issues of notification fatigue by seeking to provide notifications only when said notifications are needed, when a person has the opportunity to act on them, and is receptive to receiving support. In addition, suggested daily step goals will also be varied systematically across time. A suggested step goal will vary between a person's median steps/day, calculated from the person's previous activity measured via Fitbit, up to 3,000 steps above their median reference. The goals will continue to get progressively more difficult if a person meets their suggested step goals. The system will stop increasing suggested step goals if a person achieves a median of 12,000 steps/day as their reference. During the study, participants will wear a Fitbit for the duration to measure PA and also fill out ecological momentary assessment surveys of psychological constructs hypothesized to be key variables for the targeted dynamical computational models.
After study completion, dynamical modeling analyses appropriate for system identification will be conducted for each participant (see references for more details on the types of analyses that will be conducted). The goal is to estimate and validate the dynamical computational models, with a particular benchmark used on the degree to which a dynamical model can predict, prospectively, each person's future steps/day and response to a particular bout notification. Results from this dynamical systems modeling will then enable the development of a multi-timescale model-predictive controller driven JITAI designed to provide support for increasing walking among healthy adults, which can then be tested in a future clinical trial.
Study Design
Arms and Interventions
| Arm | Intervention/Treatment |
|---|---|
| Experimental: System Identification All participants in the study will go through a system identification experiment everyday for 270 days. |
Behavioral: System identification experiment for physical activity
The system identification experiment in Just Walk JITAI study has two key components that are the focus of the system identification experiment: daily adaptive step goal recommendations and within-day suggestions to either plan a bout of walking or to inspire reflection and, by extension, an increased urge to go for a walk.
|
Outcome Measures
Primary Outcome Measures
- Steps/day [Everyday from baseline to the end of study (for 270 days)]
This will be measured continuously for the duration of the study via a Fitbit Versa, a wrist-worn, consumer-level activity tracker.
Secondary Outcome Measures
- Behavioral Intentions [Everyday from baseline to the end of study (for 270 days)]
After each intervention notification (i.e., suggested daily step goals and bout notifications), participants will be asked to respond on their intentions to achieve the suggested target (no, maybe, yes).
- Intervention adherence [Everyday from baseline to the end of study (for 270 days)]
Adherence will be gleaned from the Fitbit and from smartphone app in terms of the degree to which persons engage with the Fitbit, use of the app, reads prompts, and responds to prompts.
- Psychological mediators/process variables [Everyday from baseline to the end of study (for 270 days)]
self-efficacy, motivation, behavioral outcomes, and internal cues to action will be measured as key process variables hypothesized to influence a person's drive and capacity to engage in walking. These will be measured daily using ecological momentary assessment questions asked within the smartphone app.
- Environmental context [Everyday from baseline to the end of study (for 270 days)]
Barriers, facilitators, schedule, and other factors that might support or hinder a person's capacity to walk will be measured daily using ecological momentary assessment.
- Min/week moderate-to-vigorous physical activity [Everyday from baseline to the end of study (for 270 days)]
This will be measured continuously for the duration of the study using the Fitbit Versa, a wrist-worn, consumer-level activity tracker.
Eligibility Criteria
Criteria
Inclusion Criteria:
-
inactive: engaging in less than 60 min/week of self-reported moderate intensity physical activity
-
adults: aged 21 or older
-
own a smartphone that can run HeartSteps (iOS or Android)
Exclusion Criteria:
-
not proficient in English, or
-
indicate medical problems that preclude physical activity as defined using physical activity readiness questionnaire (PAR-Q)
Contacts and Locations
Locations
| Site | City | State | Country | Postal Code | |
|---|---|---|---|---|---|
| 1 | University of California San Diego | San Diego | California | United States | 92093 |
Sponsors and Collaborators
- University of California, San Diego
- National Library of Medicine (NLM)
Investigators
- Principal Investigator: Eric Hekler, PhD, University of California, San Diego
Study Documents (Full-Text)
None provided.More Information
Publications
- Hekler EB, Rivera DE, Martin CA, Phatak SS, Freigoun MT, Korinek E, Klasnja P, Adams MA, Buman MP. Tutorial for Using Control Systems Engineering to Optimize Adaptive Mobile Health Interventions. J Med Internet Res. 2018 Jun 28;20(6):e214. doi: 10.2196/jmir.8622.
- Korinek EV, Phatak SS, Martin CA, Freigoun MT, Rivera DE, Adams MA, Klasnja P, Buman MP, Hekler EB. Adaptive step goals and rewards: a longitudinal growth model of daily steps for a smartphone-based walking intervention. J Behav Med. 2018 Feb;41(1):74-86. doi: 10.1007/s10865-017-9878-3. Epub 2017 Sep 16.
- Phatak SS, Freigoun MT, MartÃn CA, Rivera DE, Korinek EV, Adams MA, Buman MP, Klasnja P, Hekler EB. Modeling individual differences: A case study of the application of system identification for personalizing a physical activity intervention. J Biomed Inform. 2018 Mar;79:82-97. doi: 10.1016/j.jbi.2018.01.010. Epub 2018 Feb 1.
- 800132
- 5R01LM013107