Abstract: Objective To address the high disability and recurrence rates of stroke and the spatiotemporal limitations of traditional rehabilitation models by integrating the phased behavioral intervention model of the Health Action Process Approach （HAPA） theory with artificial intelligence （AI） technology, and investigating the application effect of a Wechat App, based on the HAPA theory, in continuous rehabilitation training for stroke patients. Methods An intelligent rehabilitation system covering the “hospital-home-community” continuum was constructed. Guided by the stage-based intervention logic of HAPA covering motivation arousal, behavioral planning, action initiation and behavioral maintenance, this study integrated AI technology and WeChat ecosystem to provide patients with whole-process, personalized rehabilitation support. Thirty stroke patients admitted to Shanghai Yangzhi Rehabilitation Hospital and Zhongshan Sub-district Health Service Center from January to April 2025 were selected and randomly divided into a control group （Group A） and an intervention group （Group B）, with 15 patients in each group. Group A received routine stroke rehabilitation orders, while Group B received AI-based remote rehabilitation training intervention via an App based on the HAPA model, in addition to routine orders. Hand joint range of motion, hand function, National Institutes of Health Stroke Scale （NIHSS） score, Berg Balance Scale score, Fugl-Meyer Assessment score, modified Barthel Index, and Carr-Shepherd score were compared between the two groups before and 3 months after the intervention. Results After the intervention, Group B showed significantly better outcomes than Group A in multiple indicators： lower NIHSS score （18.22±4.15 vs 30.46±5.23, P<0.05）, and higher Berg Balance Scale score （47.26±5.71 vs 36.28±6.42）, Fugl-Meyer Assessment score （38.02±7.18 vs 20.14±5.89）, modified Barthel Index （35.26±6.54 vs 25.45±5.77）, and Carr-Shepherd score （82.73±8.25 vs 64.35±9.11） （all P<0.05）. Furthermore, Group B demonstrated better recovery of hand joint range of motion, greater improvement in hand function, and a higher exercise compliance rate （73.3% vs 53.3%, P<0.05）. Conclusion The Wechat App, based on the HAPA theory, has been shown to effectively enhance the compliance of rehabilitation training in stroke patients, resulting in significant improvements in neurological function, limb motor function, balance ability, and activities of daily living. It provides feasible technical support for high-quality continuous rehabilitation.

Key words: Wechat platform, Stroke, Continuous rehabilitation, Health Action Process Approach, Digital rehabilitation