Exploring the Feasibility of Digital Voice Assistants for Delivery of a Home-Based Exercise Intervention in Older Adults With Obesity and Type 2 Diabetes Mellitus: Randomized Controlled Trial

Background: Current clinical guidelines for the management of type 2 diabetes mellitus (T2DM) in older adults recommendthe use of antihyperglycemic medications, monitoring of blood glucose levels, regular exercise, and a healthy diet to improveglycemic control and reduce associated comorbidities. However, adherence to traditional exercise programs is poor (<35%).Common barriers to adherence include fear of hypoglycemia and the need for blood glucose level monitoring before exercise. Digital health strategies offer great promise for managing T2DM as they facilitate patient-practitioner communication, support self-management, and improve access to health care services for underserved populations. We have developed a novel web-based software program allowing practitioners to create tailored interventions and deliver them to patients via digital voice assistants(DVAs) in their own homes. Objective: We aim to evaluate the feasibility of a 12-week, home-based, personalized lifestyle intervention delivered and monitored by DVAs for older adults with obesity and T2DM. Methods: In total, 50 older adults with obesity aged 50-75 years with oral hypoglycemic agent-treated T2DM were randomized to the intervention (DVA, n=25) or a control group (n=25). Participants allocated to the DVA group were prescribed a home-based muscle strengthening exercise program (similar to 20- to 30-min sessions) and healthy eating intervention, delivered via DVAs (AlexaEcho Show 8; Amazon) using newly developed software ("Buddy Link"; Great Australian Pty Ltd). Control group participantsreceived generalized physical activity information via email. Outcomes were feasibility, DVA usability (System Usability Scale),and objectively assessed physical activity and sedentary time (wrist-worn accelerometers). Results: In total, 45 (90%) out of 50 participants completed this study. Mean adherence to prescribed exercise was 85% (SD43%) with no intervention-related adverse events. System usability was rated above average (70.4, SD 16.9 out of 100). Compared with controls, the DVA group significantly decreased sedentary time (mean difference -67, SD 23; 95% CI -113 to -21 min/d),which was represented by a medium to large effect size (d=-0.6). Conclusions: A home-based lifestyle intervention delivered and monitored by health professionals using DVAs was feasible for reducing sedentary behavior and increasing moderate-intensity activity in older adults with obesity and T2D.