PERFORMANCE OF AI-GENERATED DRUG–DRUG INTERACTION ALERTS VERSUS PHARMACIST ASSESSMENT: A SYSTEMATIC REVIEW

Seerat Shahzad; Ayesha Afzaal; Rida Afzaal; Asma Ashraf; Hafiz Muhammad Bilal; Syed Hamid Hussain Zaidi; Abdul Rehman Abid; Zarfshan Shahzad

doi:10.22270/ujpr.v11i3.1575

Seerat Shahzad Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan.
Ayesha Afzaal Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan.
Rida Afzaal Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan.
Asma Ashraf Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan
Hafiz Muhammad Bilal Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan.
Syed Hamid Hussain Zaidi Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan.
Abdul Rehman Abid Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Lahore University of Biological and Applied Sciences, Lahore, Pakistan.
Zarfshan Shahzad Department of English, Faculty of Languages, National University of Modern Languages, Islamabad, Pakistan.

10.22270/ujpr.v11i3.1575

Keywords:

Artificial intelligence, clinical decision support, clinical pharmacy, drug-drug interactions, pharmacist, systematic review

Abstract

Background: Artificial Intelligence (AI) is increasingly being used in the medicine administration and drug–drug interaction (DDI) screening domain in healthcare. However, its reliability in comparison with the pharmacist review is unknown. This systematic review aimed to compare the effectiveness of DDI alerts generated by AI to DDI alerts generated by pharmacists.

Methodology: This review was done in accordance with PRISMA 2020 guidelines and registered with PROSPERO (CRD420251153581). The publications since 2022 were obtained from PubMed, Embase, Scopus, Web of Science, Cochrane Library, IEEE Xplore, and arXiv. The Mixed Methods Appraisal Tool (MMAT) 2018 was used to evaluate the quality of the studies.

Results: Fourteen studies of AI systems like ChatGPT, Gemini, Bing AI, and Claude were selected. AI models performed well in general drug information retrieval and were found to have significant shortcomings when it comes to clinical DDI screening. Low accuracy, inconsistent responses, severity and onset not assessed and clinically important interaction not included were common problems. Pharmacist review and validated databases like Lexicomp and Micromedex were more reliable to assess DDI.

Conclusion: AI systems can provide support for drug information and may be suitable but are currently not sufficient for autonomous drug information screening. However, pharmacist oversight will remain essential to patient safety, and additional studies are required to develop customizable AI systems for pharmacies that incorporate trusted clinical decision support tools.

Peer Review History:

Received 9 April 2026; Reviewed 11 May 2026; Accepted 13 June; Available online 15 July 2026

Academic Editor: Dr. Ahmad Najib, Universitas Muslim Indonesia, Indonesia, [email protected]

Reviewers:

Dr. Esther Marguerite Chase DJANGA, Faculty of Medicine and Biomedical Sciences. Department of Public Health. University of Yaoundé I, Cameroon. [email protected]

Dr. Dennis Amaechi, MrsFoluBabade Mini Estate , Flat 5 by Old Soldiers Quarter, Sabongari/Bwari, Abuja- Federal Capital Territory, Nigeria. [email protected]