Virtual Assistant for Exercise Adherence in Older ADULTS

The objective of this study is to determine whether a virtual assistant-controlled, home-based exercise program can improve adherence to therapeutic exercise among older adults, compared with a conventional municipal program of similar dosage. Although exercise is effective for preserving function and quality of life during aging, adherence remains suboptimal. Virtual assistants may offer a scalable and acceptable solution to support sustained participation. The study will employ a nonrandomized controlled clinical trial (quasi-experimental design) involving community-dwelling adults aged ≥60 years. Participants will be allocated to groups based on program pathway: Experimental group (Virtual Assistant/BOT): a structured home-based exercise program (60 minutes per session, 3 sessions per week for 12 weeks), guided and monitored by a virtual assistant (BOT), including automated session delivery, reminders, and adherence tracking. Control group (Municipal program): older adults already participating in Enguera se saludable, an onsite, instructor-led municipal program with equivalent frequency and duration (60 minutes, 3 times per week for 12 weeks). Attendance records from the municipal program will quantify adherence. The primary outcome is exercise adherence, defined as the proportion of prescribed sessions completed during the 12-week period, with predefined adherence thresholds (e.g., ≥75% of sessions completed). Secondary outcomes include changes in muscle strength, balance, functional performance, and health-related quality of life. Additional measures will include user perceptions, usability, and satisfaction with the virtual assistant to evaluate acceptability and feasibility. Given the nonrandomized design and the pre-existing enrollment of the control group, the study will incorporate the following methodological safeguards: Baseline demographic and clinical characteristics (e.g., age, sex, comorbidities, prior activity level) will be comprehensively documented. Baseline imbalances will be controlled through statistical adjustment (e.g., covariate adjustment, propensity score procedures, and sensitivity analyses) to reduce selection bias and confounding. Exercise dosage (frequency, duration, and intervention length) will be matched across groups, or, when small discrepancies arise, dosage will be included as a covariate in the analyses. This study is expected to contribute evidence regarding whether a virtual assistant-based exercise program represents a viable, effective, and acceptable alternative to conventional program delivery models for promoting active and healthy aging in community-dwelling older adults.