Artificial intelligence and rule-based decision support for medication optimization in older adults: a systematic scoping review of polypharmacy management tools and ethical challenges

Polypharmacy—commonly defined as the concurrent use of five or more medications—is highly prevalent in older adults and increases the risk of potentially inappropriate medications (PIMs), drug–drug interactions (DDIs), adverse drug events (ADEs), and functional decline. This PRISMA-ScR–guided scoping review synthesised evidence from 81 studies on both artificial intelligence (AI) and established rule-based clinical decision support systems (CDSS) for medication optimisation in adults aged ≥ 65 years with polypharmacy. Both AI approaches—including machine learning (ML), deep learning (DL), and natural language processing (NLP)—and traditional rule-based tools such as MedSafer and STOPP/START were included to capture the full landscape of digital prescribing support. Evidence shows consistent process-level improvements such as more accurate PIM identification, medication reconciliation, and adherence monitoring, yet patient-level outcomes (ADE reduction, hospitalisation, mortality) remain limited and heterogeneous. Persistent gaps include under-representation of frail and cognitively impaired populations, limited external validation, and minimal age- or frailty-stratified performance reporting. A full list of included studies is provided in Supplementary Table S1. Broader validation, age-inclusive datasets, integration of social determinants of health, and ethically governed implementation are essential to ensure AI and rule-based CDSS deliver equitable, meaningful improvements in geriatric pharmacotherapy.