(073) Effects of AI-Assisted Drug Therapy Management on Medication-Related Issues in Elderly Patients: A Systematic Review

Background: Hospitalized elderly patients frequently encounter drug-related problems (DRPs), leading to extended hospitalizations, deteriorated health conditions, higher medical expenses, and mortality. Utilizing artificial intelligence (AI) for medication management may enhance drug therapy optimization and reduce DRPs among these patients.

Objectives: This systematic review assesses the efficacy of AI-assisted strategies in refining drug therapy management and mitigating DRPs for elderly patients in hospital settings.

Methods: An extensive search was carried out in databases such as PubMed, Scopus, and Google Scholar for articles published between 2012-2022. Studies evaluating computerized interventions, such as electronic prescribing, computerized physician order entry, and clinical decision support systems, involved in improving the appropriateness of care for elderly patients in hospital settings were included. Three reviewers independently screened titles, abstracts, and full texts. Data extraction included study details, intervention, outcomes, drug-related problem types, and recommendation acceptance rates. Risk of bias and methodological quality were assessed using standard tools.

Results: Out of 567 studies found, 194 were removed due to duplication. Through further screening for relevance and eligibility, 13 studies were selected for review, including five randomized controlled trials, two non-randomized trials, and six cohort studies, with quality scores between 79% and 100%. These studies focused on clinical decision support systems and their influence on factors like inappropriate medication use, medication suitability, adverse drug reactions. Primary outcomes showed 20-30% reductions in PIMs, 7.3% absolute risk reduction in adverse drug reactions, and significant improvements in medication appropriateness scores. Secondary outcomes revealed 4.7-54.8% recommendation acceptance rates, non-significant decreases in length of stay (8 vs. 9 days) and mortality (4-5%). Overall, computerized interventions effectively identified drug-related problems, reduced PIMs and adverse events, and moderately improved prescribing appropriateness in elderly hospital inpatients. Although AI applications showed promise in bettering prescription practices, conclusive impacts on major health outcomes were not observed. Challenges such as alert fatigue and non-adherence to AI recommendations were noted as significant obstacles.

Conclusions: AI-driven strategies show promise in enhancing medication safety and optimizing treatment plans for elderly hospital patients. Yet, further studies are necessary to confirm their impact on health outcomes, including healthcare usage, disease burden, and mortality rates.