A Digital Tool to Help Doctors and Pharmacists Prescribe Medicines More Safely for Older Adults in Oman (POLYMED)

Study Rationale and Design Framework Geriatric prescribing safety represents a critical health services focus. This protocol evaluates an investigator-assigned digital health intervention designed to systematically intercept prescribing risks at the point of care. While the protocol includes a baseline run-in period to parameterize the software, the study functions globally as a prospective interventional trial. It measures the clinical efficacy of an AI-assisted clinical decision support system (CDSS) within an active primary care workflow.

Phase 1: Baseline Assessment & Parameterization The study initiates with a prospective baseline assessment of a primary care patient cohort aged 65 or older. Investigators systematically extract data regarding active medication regimens, clinical diagnoses, and health expenditures. This phase applies the 2023 AGS Beers Criteria and STOPP/START v3 criteria to establish a strict pre-intervention standard-of-care baseline for polypharmacy prevalence, PIP rates, and associated medication costs.

Phase 2: Intervention Engineering & System Validation Data derived from Phase 1 are immediately used to program, calibrate, and validate the proprietary OPIP Score engine. This engine drives the bilingual (Arabic/English) POLYMED web calculator. The software applies machine learning algorithms to generate an explainable-AI narrative. This narrative visually highlights specific pharmacological risk vectors to actively prompt clinicians during patient evaluations.

Phase 3: Prospective Interventional Workflow The core interventional component of the protocol utilizes a prospective before-and-after design to evaluate the tool's clinical impact.

1. Protocol-Mandated Assignment: The investigator actively assigns participating primary care clinicians to integrate the POLYMED calculator into their clinical consultations. For all prospectively enrolled participants in this phase, clinicians are required by the study protocol to execute the web calculator at the point of care during the patient visit.
2. Clinical Action and Decision Support: The software functions as an active advisory system. Upon receiving the real-time, explainable-AI risk alerts, clinicians are required by the study workflow to conduct a structured medication review. While clinicians maintain complete medical autonomy over final prescribing modifications, the review process itself is forced by the trial protocol.
3. Justification for Interventional Framework: This study is classified as interventional because the systematic application of this specific AI-driven risk calculator is explicitly mandated by the investigator's protocol to alter clinical behavior and evaluate its direct effect on patient health outcomes. This intensive workflow and real-time risk feedback loop are not part of routine medical care in Oman and would not occur outside the context of this clinical trial. The study evaluates the resulting differences in prescribing patterns, safety metrics, and medication costs between the baseline standard of care and the active intervention phase.