A Randomized Controlled Trial of a Multi-domain Smartphone Intervention Incorporating Generative AI for Healthy Older Adults

Efficacy and Scalability of a Multi-domain Digital Health Intervention Via Smartphone Application and Wearable Data: A Randomized Controlled Trial in Healthy Elderly Population

This study evaluates the efficacy of a smartphone-based multi-domain intervention designed to promote healthy aging and prevent cognitive decline in older adults. Building on preliminary findings that automated alerts improved exercise, sleep and nutrition-thereby mediating enhanced cognitive performance-we developed a dedicated application that integrates longitudinal Fitbit data (activity, sleep, and nutrition) with daily interactive dialogues powered by Generative AI (LLM).

A six-month Randomized Controlled Trial (RCT) will be conducted with 300 healthy participants aged 65 and older. The primary objective is to analyze how these digital interventions influence health outcomes, particularly cognitive and motor functions. Furthermore, we aim to construct a predictive mathematical model to estimate health states based on wearable data streams. Finally, the study will assess the potential for scalability to facilitate the large-scale social implementation of this digital health service.

Inclusion Criteria:

• Able to use a smartphone and wearable device (Fitbit).

Exclusion Criteria:

• Physical conditions that prohibit regular exercise. Individuals diagnosed with dementia