Sulaiman Dilveen M, Shaba Suhail S, Almufty Hind B, Sulaiman Asmaa M, Merza Muayad A
Department of Pharmacology, College of Pharmacy, University of Duhok, Duhok, IRQ.
Department of Pharmaceutics, College of Pharmacy, University of Duhok, Duhok, IRQ.
Cureus. 2023 Sep 9;15(9):e44961. doi: 10.7759/cureus.44961. eCollection 2023 Sep.
Aim This study aimed to critically appraise the drug-drug interaction (DDI) screening performance of Google Bard (Google AI, Mountain View, California, United States) by comparing it with the authorized Lexicomp® Online™ database (Wolters Kluwer Health, Philadelphia, Pennsylvania, United States). Methods This cross-sectional study was conducted between April 2023 and August 2023, and enrolled 414 prescriptions that had been collected randomly between April 2023 and June 2023. These prescriptions were processed individually by Lexicomp online and Google Bard to screen for DDIs between antimicrobials and other prescribed medications. Results The total number of DDIs based on Lexicomp and Google Bard were 90 and 68, respectively. Cohen's Kappa (κ) values showed that there was a nil to slight agreement between Lexicomp and Google Bard regarding the DDI risk rating (κ=0.01). Regarding the severity rate, there was a slight agreement between them (κ=0.02), but in terms of reliability rate, there was no agreement (κ =-0.02). Conclusion This study unveiled differences between Lexicomp and Google Bard regarding their DDI identification, severity rating, and reliability rates. It is fundamental to consider that both tools have their strengths and weaknesses and, therefore, should not be individually depended on for final clinical decisions. However, Lexicomp can be considered authoritative in screening DDIs, but Google Bard currently lacks the necessary precision and reliability for conducting such screenings.
目的 本研究旨在通过将谷歌巴德（Google Bard，谷歌人工智能，美国加利福尼亚州山景城）与经授权的Lexicomp® Online™数据库（美国宾夕法尼亚州费城的威科医疗集团）进行比较，对其药物相互作用（DDI）筛查性能进行批判性评估。方法 本横断面研究于2023年4月至2023年8月进行，纳入了2023年4月至2023年6月期间随机收集的414份处方。这些处方分别由Lexicomp在线平台和谷歌巴德进行处理，以筛查抗菌药物与其他处方药物之间的药物相互作用。结果 基于Lexicomp和谷歌巴德筛查出的药物相互作用总数分别为90例和68例。科恩kappa（κ）值表明，Lexicomp和谷歌巴德在药物相互作用风险评级方面的一致性为零至轻微（κ=0.01）。在严重率方面，两者之间存在轻微一致性（κ=0.02），但在可靠性率方面，不存在一致性（κ=-0.02）。结论 本研究揭示了Lexicomp和谷歌巴德在药物相互作用识别、严重程度评级和可靠性率方面存在差异。必须认识到这两种工具都有其优缺点，因此，最终临床决策不应仅依赖于它们。然而，Lexicomp在筛查药物相互作用方面可被视为权威，但谷歌巴德目前在进行此类筛查时缺乏必要的准确性和可靠性。
