ChatGPT -Based Intervention for Social Frailty in Older Women With CHF : Gender Differences, Predictive Modeling, and Longitudinal Evaluation

The study is planned to be completed over three years and will focus on 500 hospitalized older adults with a diagnosis of CHF. The Biopsychosocial model will be used to conduct this study. Phase 1: This phase aims to determine the incidence rates of sarcopenia, frailty/social frailty among hospitalized older adults, and to compare these outcomes in gender differences. Sarcopenia will be identified according to the AWGS guidelines, while clinical characteristics, including ejection fraction, stage of CHF, and comorbidities, will be collected via chart review using the 2024 Guidelines of the Taiwan Society of Cardiology(Y. H. Li et al., 2024). Frailty and social frailty will be measured using the CFS, Tilburg Frailty Indicator (TFI), and Makizako's Social Frailty Questionnaire, respectively, and physical activity levels will be assessed with the International Physical Activity Questionnaire - Short Form (IPAQ-S). To develop a predictive model examining risk factors for social frailty and health outcomes, with a focus on gender differences, we will employ Structural Equation Modeling (SEM). SEM allows for the analysis of complex relationships among variables, facilitating a comprehensive understanding of how these factors interact differently across genders. Phase 2: A quasi-experimental design will be implemented, involving 174 hospitalized older adults with low social frailty scores. Participants will be randomly assigned to one of three groups. Group 1 will use the ChatGPT app for 15-30 minutes daily, with the app actively monitoring their emotional status and automatically notifying the research team and family members if signs of negative mood are detected. Group 2 will be encouraged to increase their social participation through community-based activities. The Control Group will continue their routine lifestyle without additional interventions. Before the intervention, one-on-one training will be provided to Group 1 participants on how to use ChatGPT. Key outcome measures, including mood status (loneliness, anxiety, and depression), physical activity (assessed via IPAQ-S), overall frailty (CFS \& TFI), social frailty scores, and other health indicators, will be assessed at baseline and immediately post-intervention. Phase 3: The aim of this phase is to monitor and compare the longitudinal health outcomes of interventions among the three groups. Key health outcomes will be assessed and compared every three months over a one-year period.