Towards Proactive Medication Safety: Artificial Intelligence-Based Prediction, Detection, and Prevention of Drug-Related Problems in Clinical Practice

Abstract Background Medication-related problems, including adverse drug reactions (ADRs), adverse drug events (ADEs), drug–drug interactions (DDIs), medication errors, and inappropriate antimicrobial use, remain important challenges in clinical practice. Artificial intelligence (AI) has developed rapidly and may help healthcare professionals identify medication-related risks earlier. Previous studies have reported the use of machine learning, deep learning, natural language processing, and newer generative AI methods in pharmacovigilance, drug-interaction prediction, clinical decision support, and antimicrobial stewardship. Objective This study aimed to review AI's role in predicting, detecting, and preventing drug-related problems and to compare its applications across pharmacovigilance, drug–drug interaction prediction, clinical decision support, medication safety, and antimicrobial stewardship. Methods We conducted a literature review of 39 selected articles on AI and medication safety. The literature was organised into major areas, including ADR/ADE prediction and pharmacovigilance, DDI prediction, clinical decision support, generative AI and EHR-based medication safety applications, and antimicrobial stewardship. The findings were compared according to AI methods, clinical purpose, safety application, benefits, and major limitations. Results The reviewed literature shows that AI can analyse large amounts of clinical and medication data to identify patients at risk of ADRs and ADEs, detect possible DDIs, support medication decisions, and improve pharmacovigilance. Machine learning and deep learning methods have shown promising results in prediction and classification tasks. AI has also been used to support antibiotic selection, antimicrobial resistance prediction, and antimicrobial stewardship. Generative AI and large language models are emerging in DDI identification, clinical decision support, and pharmacovigilance. However, most studies remain retrospective, and external validation and prospective clinical testing are still limited. Conclusion AI has strong potential to support a proactive medication-safety system by predicting risks, detecting problems early, and supporting preventive action. However, AI should support rather than replace healthcare professionals. Future research should focus on high-quality data, external validation, explainable AI, patient privacy, clinical integration, and real-world testing. The ultimate goal should be to develop safe, reliable, and patient-centred AI systems that help reduce medication-related harm and improve clinical outcomes. Keywords: Artificial Intelligence; Medication Safety; Drug-Related Problems; Pharmacovigilance; Adverse Drug Reactions; Drug–Drug Interactions; Clinical Decision Support Systems; Antimicrobial Stewardship