Skip Navigation
Skip to contents

JPMPH : Journal of Preventive Medicine and Public Health

OPEN ACCESS
SEARCH
Search

Search

Page Path
HOME > Search
1 "Meta AI (Llama 4 Scout)"
Filter
Filter
Article category
Publication year
Original Article
Agreement Between Artificial Intelligence Platforms and the UpToDate Lexidrug Database in Identifying Antihypertensive Drug–drug Interactions in Older Adults
Januarius Tomi Kiik, Agung Endro Nugroho, Soni Siswanto
J Prev Med Public Health. 2026;59(5):531-541.   Published online August 24, 2026
DOI: https://doi.org/10.3961/jpmph.26.359
  • 14,118 View
  • 647 Download
AbstractAbstract AbstractSummary PDFSupplementary Material
Objectives
This study aimed to compare the agreement and consistency of classifications generated by ChatGPT-4.5 and Meta AI (Llama 4 Scout) with those of the UpToDate Lexidrug reference standard for identifying potential drug interactions involving antihypertensive medications in older adults.
Methods
A cross-sectional study was conducted using electronic medical record data from January 2024 to December 2024. Sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) were calculated to assess agreement between classifications generated by ChatGPT-4.5 or Meta AI and those of UpToDate Lexidrug as the reference standard. In addition, Cohen’s kappa values were used to assess inter-rater agreement regarding the severity of the identified drug interactions.
Results
ChatGPT-4.5 showed greater agreement with the reference standard (UpToDate Lexidrug), with an accuracy of 80.5%, sensitivity of 86.2%, specificity of 78.0%, PPV of 62.8%, and NPV of 92.9%, compared with Meta AI, which had an Should be 70.6%, sensitivity of 75.0%, specificity of 68.7%, PPV of 51.0%, and NPV of 86.3%. Cohen’s kappa analysis showed fair agreement between both platforms and the reference standard (0.367 and 0.359, respectively).
Conclusions
ChatGPT-4.5 demonstrated greater alignment with the UpToDate Lexidrug classification compared to Meta AI across all evaluated metrics. However, both artificial intelligence platforms showed only fair agreement in severity classification, indicating substantial differences between the classifications generated by the platforms and those of the reference database.
Summary
Key Message
In this study of 819 unique antihypertensive drug pairs in older adults with hypertension, ChatGPT-4.5 showed higher sensitivity, specificity, and overall accuracy in identifying drug–drug interactions compared to Meta AI (Llama 4 Scout). But both ChatGPT-4.5 and Meta AI showed only fair agreement with the reference database for interaction-severity classification. These results support the use of generative AI as an adjunctive screening tool, not a replacement for validated drug-interaction databases or clinical judgment.

JPMPH : Journal of Preventive Medicine and Public Health
TOP