Informatics-Based Digital Application to Promote Safe Exercise in Middle-Aged Adults with Type 1 Diabetes

ABSTRACT
Type 1 diabetes (T1D) affects ~1 million American adults and increases the risk of mortality attributable to
cardiovascular disease by 800%. Current evidence-based T1D self-management interventions target glycemic
control but ignore other modifiable health concerns prevalent in T1D such as hypertension and obesity. Exercise
interventions could provide a novel solution if they could innovatively address the diabetes management and
psychosocial challenges around exercise posed by T1D. Continuous glucose monitoring (CGM) allows patients
and providers to comprehensively track the short- and long-term outcomes of exercise. Evidence-based
interventions to translate CGM technology into sustainable adherence to exercise-related behaviors are lacking.
Our human-delivered pilot intervention provided previously sedentary adults with T1D access to exercise videos
and monthly client-centered discussions of their CGM and exercise data with an exercise coach. Participants
said these improved exercise management behavioral skills and motivation, but only transiently. They stated a
need for more frequent and sustained contact, requiring automated mobile tools that this proposal will develop.
These tools include just-in-time adaptive text messages to overcome exercise barriers at times of vulnerability,
weekly personalized reviews of short-term exercise safety hazards with tips to avoid them, and monthly
personalized evaluation of long-term impact of exercise on blood glucose levels via Bayesian modeling. The
program represents stage 1 of the NIH intervention development model: intervention generation, refinement,
modification, adaptation. These steps will be accomplished by a feasibility study evaluating user satisfaction and
mathematical robustness of an alpha version, using these results to modify the alpha version into a beta version,
and then testing the beta version in a nonrandomized crossover clinical trial. Lastly, the databank of
biobehavioral metrics generated by this trial (exercise, CGM, mood and sleep diaries for ~ 7,000 person-days)
will be subjected to dimensionality reduction to identify biobehavioral subtypes of baseline and early intervention
data. We will test whether these subtypes help predict longer-term intervention response and/or flag specific
biobehavioral feature combinations that drive intervention responsiveness. These findings will lay a foundation
for Dr. Ash’s future work developing precision medicine approaches. Alongside this research Dr. Ash will
complete training in the domains of 1) diabetes management and technology; 2) mobile health (mHealth)
intervention development; and 3) dimensionality reduction analytics. The training plan includes a strategic
combination of mentor-led trainings, coursework, grant writing, and attendance at relevant conferences and
workshops. Dr. Ash has assembled a mentoring team in T1D self-management and technology, multiple health
behavior change intervention development, mHealth development, and informatics. FitscriptLLC and PiLR Health
will provide customized intervention tools and data capture software. Dr. Gerstein’s laboratory will support data
storage, processing, and analytics.

Grant Number: 5K01DK129441-04
NIH Institute/Center: NIH
Principal Investigator: Garrett Ash