Evaluating the Role of AI-Based Clinical Decision Support Systems in Reducing Medication Errors

Abstract

Medication errors remain a significant challenge in clinical practice, contributing to patient morbidity, increased healthcare costs, and adverse outcomes. Clinical Decision Support Systems (CDSS) have been developed to mitigate such errors, with recent advances in artificial intelligence (AI) enabling predictive and adaptive interventions. This study evaluates the role of AI-based CDSS in reducing medication errors, comparing their performance to traditional rule-based systems while examining factors such as patient demographics, comorbidities, alert acceptance, and medication type. A retrospective dataset of 300 prescriptions was simulated, encompassing patient age, comorbidities, medication type, CDSS type, alert generation, alert acceptance, and error occurrence. Analyses included descriptive statistics, error rate calculations, severity distribution, age-stratified error assessment, and cumulative alert tracking. Comparative evaluation between AI-based and rule-based CDSS focused on overall error reduction, alert acceptance, and high-risk medication identification. Key performance metrics such as medication error rate, alert acceptance rate, and severity distribution were computed to quantify system effectiveness. Results demonstrated that AI-based CDSS reduced overall medication errors to 25.3%, compared to 34.7% in rule-based systems. High-risk medications such as anticoagulants and cardiac drugs exhibited error rates of 41.7% and 33.3%, respectively, with AI intervention decreasing both moderate and high-severity errors. Alert acceptance for prescriptions with errors reached 64%, indicating effective clinician engagement. Age-stratified analysis revealed error rates increased from 20% in the 18–30 age group to 42.7% in patients aged 71–90, while patients with higher comorbidity counts exhibited greater error variability. The study concludes that AI-enabled CDSS effectively reduces medication errors, particularly for high-risk drugs, elderly patients, and complex comorbidity profiles. Implementation of predictive and adaptive decision support enhances medication safety, improves alert relevance, and mitigates high-severity errors, underscoring the potential of AIdriven systems to strengthen clinical workflows and patient outcomes.

Ähnliche Arbeiten