March 7, 2025 Brainstorm Overview
# Improving the App, Interventions, and/or Study Design for the Stroke Rehabilitation Micro-Randomized Trial
Collaboration with Prabhat Pathak, James Arnold, Pat Puma, and Conor Walsh of the [Harvard Biodesign Lab](https://biodesign.seas.harvard.edu/lab-members); J. P. Tuttle (Software Engineer) and Paul Sabin (Director) of the [Harvard Move Lab](https://www.movelab.seas.harvard.edu/); and Lynn Blaney (Occupational Therapist) and David Lin (Neurorehabilitation Physician Scientist) of the [Laboratory for Translational Neurorecovery](https://www.massgeneral.org/neurology/research/david-lin-lab-translational-neurorecovery) at Massachusetts General Hospital

## Goals for the Brainstorm
Our goals for this brainstorm are to evaluate and improve our (1) app's design and functionality, (2) the proposed intervention components, and (3) the proposed study design for the planned micro-randomized trial (MRT). Our MRT investigates digital health support for [chronic stroke survivors](https://www.neurolutions.com/after-stroke/a-guide-to-chronic-stroke-recovery-and-symptoms/) rehabilitating their upper limb function. Our focus is on augmenting digital health support like telerehabilitation (e.g., video calls with occupational therapist) with mobile health (mHealth) interventions that encourage affected arm use during everyday life.

## Background
In our [last brainstorm back on October 4, 2024](https://hackmd.io/@cynvq9KSSpmj1RAX-aca1g/ryAa__oCC) (see Background section for review), we discussed the importance of activities of daily living (ADLs) and therapist-informed exercises for chronic stroke survivors recovering the functionality of their affected arm. Approximately [80 million individuals around the world have survived a stroke](https://www.world-stroke.org/world-stroke-day-campaign/life-after-stroke/up-again-after-stroke#:~:text=Stroke%20is%20a%20leading%20cause,is%20possible%20to%20get%20%23UpAgainAfterStroke.), and [50-70% of stroke survivors live with upper limb motor impairment post-stroke](https://doi.org/10.1161/01.STR.19.12.1497). Using one's affected arm to perform ADLs is critical not only to [improving one's quality of life](https://doi.org/10.1589/jpts.26.417) via regained independence and self-efficacy, but also has been shown to [improve motor outcomes for one's affected arm](https://doi.org/10.1136/bmj.39343.466863.55). ADLs include activities such as brushing one's teeth, combing one's hair, and tying one's shoes. Occupational therapy can help guide a stroke survivor with tailored goals and exercises (e.g., which exercises to perform at home), encouragement, and accountability -- but synchronous occupational therapy, even if provided remotely, can only account for a small fraction of a stroke survivor's life (e.g., 1-3 hours per week). Support is needed to guide and encourage affected arm use during daily life to help extend the reach of occupational therapist support.

Our proposed digital health solution will support chronic stroke survivors during daily life, guiding them, encouraging them, and helping them plan to use their affected arm for ADLs and exercises based on tailored goals and messaging from occupational therapists. We are designing an Apple Watch app called "Reachable" to encourage affected arm use during daily life (Apple Watch is used because gyroscope data is collected, independent LTE connectivity is available, and sensor quality is excellent). Reachable includes two intervention components (as of now): (1) personalized push notifications delivered throughout the day to encourage affected arm use, tailored to the long-term goals of the survivor (e.g., playing basketball with a grandchild or even snowboarding again) and (2) activity planning at the beginning of the day based on [implementation intentions](https://en.wikipedia.org/wiki/Implementation_intention), where a stroke survivor makes the intention to touch their toes three times using their affected arm, for example, each time they stand up from a chair. Reachable also records inertial measurement unit data (i.e., linear acceleration and angular velocity) to track movement of the wrist (and ultimately inform support using sensor feedback in the future). To optimize Reachable and investigate in which contexts the provision of personalized reminders and/or goal setting interventions improves affected arm use, we are planning to conduct a MRT (pure exploration in reinforcement learning terms) starting late 2025.

## Questions to Think About
- What are your thoughts on "Reachable," the Apple Watch app? Would you improve the design or functionality of Reachable in any way for the purposes of this study?
- Do you think the proposed interventions have a reasonable chance at encouraging affected arm use during daily life? If you received interventions of this form to encourage non-dominant arm use and you had motivation to improve your non-dominant arm's function (e.g., for sports), would these interventions work for you?
- Are there improvements to the MRT study design you think we should consider? Does the protocol timeline make sense? What are your thoughts on the decision times for each of the intervention components? What are your thoughts on the proximal outcomes? Is there anything else you think is important to do or ask during the initial study visit, post-baseline video call, mid-study video call, or final study visit?

## The App and Example Intervention Content
## MRT
### Research Questions
1. How do the personalized reminders / encouraging messages to use one's affected arm contribute to changes in affected arm use over the next hour, and how do these effects change as a function of context (previous arm usage, time of day, etc.) and length of time in the study?
2. How do the goal setting interventions arm contribute to changes in affected arm use over the next 24 hours, and how do these effects change as a function of context (e.g., day of the week) and length of time in the study?
### Study Design Summary
- In-lab session at the beginning for baseline questionnaires, functional assessments, and intervention personalization (as well as additional data collected for other research purposes)
- 1-week baseline observation with no interventions where activities are logged and data are collected (for other research purposes such as activity recognition algorithm development), followed by 5-week MRT
- Zoom call after baseline week to finalize MRT and intervention personalization
- Another Zoom call between weeks 2 and 4 of the MRT as a form of telerehabilitation and to also serve as a mid-study check-in
- In-lab session at the end for post-study questionnaires, functional assesments, and feedback (as well as additional data collected for other research purposes)
- Decision Times
- Every morning for goal setting
- Three times per day for personalized reminders
- All interventions are sent with 0.5 probability
- Apple Watches with LTE worn on both wrists to collect data from both arms
- Distal outcomes include questionnaires regarding one's self-efficacy and perceived affected arm usage, as well as functional motor assessments
- Plan is to ultimately include 30 chronic stroke survivors who are mild to high moderate functioning as participants in the full-blown study

## Pilot Studies Underway!
Our first pilot study began on February 6 and is currently underway with Prabhat Pathak as our first pilot participant. We are in week 4 of the MRT, and Prabhat should be able to attend the brainstorm, so you should have the chance to see what he has to deal with! I will weave in lessons learned from the pilot that have informed some of what we are doing, and I am happy to share other lessons learned as well if time allows and there is interest.