Abstract
BACKGROUND: Alzheimer disease (AD) affects cognition, treatment adherence, family connections, and health care resource allocation. Most patients with AD have low adherence to medication therapy due to the limitations associated with cognitive impairment. Therefore, increasing the involvement of patients and their family members in medication management is important to improve treatment outcomes and reduce the burden of care. OBJECTIVE: This study explores the potential application of artificial intelligence (AI) in medication management for Chinese patients with early- to mid-stage AD focusing on enhancing medication adherence. The study first predicts and evaluates key factors through an online Delphi study, which provides a basis for their subsequent incorporation into the AI model as input variables to enable prediction of medication-taking behaviors. Since AI research in medication management for this population is still undeveloped, this paper further explores the multiple potentials of AI from a theoretical view, including drug dosage optimization, multidrug interaction detection, and family education support. It will provide a preliminary direction and theoretical basis for the development of an intelligent medication management system in the future. METHODS: The exploratory online Delphi study with no modification predicted the key factors influencing medication adherence. Based on the results, the study confirmed the potential of AI to improve adherence. Participation by 12 experts in 3 rounds systematically assessed the core elements influencing patients' adherence to their medication. RESULTS: Family care, social support, environmental factors, emotional support, and patient behaviors were identified as the primary factors influencing medication adherence among Chinese patients with AD. These factors were validated and ranked through iterative Delphi rounds, with family care and social support receiving the highest importance scores. The Wilcoxon signed-rank test indicated no significant difference between rounds (P=.06), supporting the stability of the consensus. These findings establish a foundational set of variables for AI systems that predict and enhance medication adherence. CONCLUSIONS: This study highlights the critical factors affecting medication adherence by Chinese patients with AD. It was designed as an exploratory online Delphi study to identify and prioritize key influencing factors, rather than to validate a specific AI-based system, and the findings provide a theoretical foundation for future AI-informed interventions. The results also indicate theoretical potential roles for AI in supporting medication management, such as optimizing drug dosage, detecting multidrug interactions, and enhancing family education.