From Hospitals to Home: Mapping the Evolving Preferences and Needs for Artificial Intelligence in Orthopedic Transitional Care:Cross-Sectional Study (Preprint)

BACKGROUND With the acceleration of global aging and rising demand for orthopedic surgeries, Enhanced Recovery After Surgery (ERAS) protocols have shortened hospital stays but created a "transitional care gap," shifting complex rehabilitation tasks to the home setting. While artificial intelligence (AI) offers potential solutions, patient perceptions regarding its role—ranging from informational chatbots to functional monitoring systems—remain underexplored. OBJECTIVE This study aims to map the evolution of care needs from hospital to home recovery, and to identify specific preferences and independent predictors of AI acceptance in orthopedic transitional care. METHODS A multicenter, cross-sectional survey was conducted with orthopedic patients across 33 hospitals in Guangdong, China. A total of 860 questionnaires were initially collected, and 752 valid responses were included in the final analysis after strict quality control (excluding response duration ≤ 180s). Data were collected on demographics, evolving task priorities across the care continuum, and perceived challenges based on an extended Technology Acceptance Model (TAM). The structure of perceived challenges was validated using Exploratory Factor Analysis (EFA). Descriptive mapping and multivariable logistic regression were performed to identify the "evolving preferences" and independent determinants of the willingness to use AI assistants. RESULTS Overall willingness to use AI was high (604/752, 80.3%). Patient priorities exhibited a fundamental shift from "passive compliance" (e.g., pain management, understanding instructions) during hospitalization to "active safety assurance" (e.g., fall prevention, motion correction) in the home setting. EFA identified 3 distinct challenge dimensions: Home Rehabilitation Self-Management Barriers, Lack of Professional Support, and Symptom Uncertainty. In multivariate analysis, significant predictors of AI acceptance included presence of comorbidities (adjusted Odds Ratio [aOR] 1.72, 95% CI 1.09–2.69), older age (aOR 1.02, 95% CI 1.00–1.03), and progression to later rehabilitation stages (aOR 1.28, 95% CI 1.01–1.62). CONCLUSIONS The transition from hospital to home involves a fundamental shift in patient needs from information acquisition to functional safety assurance. AI acceptance in this context is driven by a "Vulnerability Hypothesis," where older and clinically vulnerable patients actively seek digital support to overcome physical execution barriers. However, widespread adoption is currently constrained by a digital divide related to geography and family support. To be clinically effective, future orthopedic AI systems must move beyond generic chatbots to become "Hybrid Coaches"—integrating computer vision and sensor technology to provide real-time motion correction and fall prevention—thereby addressing the specific "Action Gap" that defines the transitional care period. CLINICALTRIAL This study is not a clinical trial, so trial registration is not required.