Empowering rural physicians through an AI-assisted intervention for deprescribing in older adults with comorbid diabetes and hypertension: protocol for a feasibility study.

Background: In China's primary healthcare system, older adults with both hypertension and diabetes are a key focus. These patients often deal with polypharmacy, which heightens the risk of adverse drug reactions, lowers treatment adherence, and increases medical burdens. Deprescribing, an evidence-based approach to optimizing medication regimens, is challenging for rural physicians due to barriers like limited clinical knowledge, lack of systematic decision support, and time constraints. Artificial Intelligence (AI) presents a promising solution, yet the feasibility of AI-assisted deprescribing in rural primary care is underexplored. This protocol describes a study designed to evaluate the feasibility, acceptability, and appropriateness of an AI agent for rural physicians managing older adults with these comorbidities. Methods and analysis: This protocol outlines a single-arm, pre-post feasibility study to be conducted in rural primary healthcare institutions across 11 administrative regions in Zhejiang Province, China. The study aims to recruit 30 eligible rural physicians and at least 150 older patients (aged ≥65 years) who have concurrent diagnoses of hypertension and diabetes and are taking five or more long-term medications. The intervention utilizes an AI agent with clinical medication guidance functions to assist rural physicians in the deprescribing process. The follow-up period spans 6 months. The primary outcome is the change in the total number of medications per patient after the intervention. Secondary outcomes include patient clinical outcomes and rural physicians' acceptance of the AI agent. Quantitative data will be analyzed using descriptive and inferential statistics, while qualitative data from semi-structured interviews will undergo thematic analysis to identify barriers and facilitators to implementation. Discussion: This protocol introduces the inaugural study evaluating the feasibility, acceptability, and appropriateness of a low-cost, independently developed AI agent tailored for rural physicians managing older adults with both hypertension and diabetes. It aims to generate empirical data on implementation processes, participant experiences, and contextual barriers. This evidence will directly inform the refinement of recruitment strategies, optimization of intervention protocols and AI agent functionality, and support future large-scale randomized controlled trials (RCTs). The study addresses critical practical challenges in deprescribing within resource-limited rural settings.