Skip to main content
ARS Home » Northeast Area » Wyndmoor, Pennsylvania » Eastern Regional Research Center » Microbial and Chemical Food Safety » Research » Publications at this Location » Publication #425608
Research Project: Technology Development, Evaluation and Validation for the Detection and Characterization of Chemical Contaminants in Foods

Location: Microbial and Chemical Food Safety

Title: Insights powered by artificial intelligence: Analyzing the extent of method validation in pesticide residue literature

Author
item RITER, LEAH - Bayer Cropscience
item Lehotay, Steven
item SWARTHOUT, JOHN - Bayer Cropscience

Submitted to: Journal of Agricultural and Food Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/29/2025
Publication Date: 6/6/2025
Citation: Riter, L.S., Lehotay, S.J., Swarthout, J. 2025. Insights powered by artificial intelligence: Analyzing the extent of method validation in pesticide residue literature. Journal of Agricultural and Food Chemistry. https://doi.org/10.1021/acs.jafc.5c04574.
DOI: https://doi.org/10.1021/acs.jafc.5c04574

Interpretive Summary: Artificial intelligence (AI) has recently achieved sufficient performance capabilities and availability for evaluation in a variety of purposes. Data extraction of desired information from the scientific literature is one example of an arduous, time-consuming task by humans that AI could manage more quickly and easily. However, any new use of AI needs to be conceptually defined, prompts developed, and outcomes validated through performance testing. This invited perspectives article describes the process and author experiences using AI for the first time in a literature extraction task to assess trends in the reported validation parameters from a subset of publications involving mass spectrometry. The main conclusion was the AI more efficiently achieved the same quality of results as human experts in the field, but experience and care is needed to define and refine prompts to yield acceptable accuracy.

Technical Abstract: Artificial intelligence (AI) has recently achieved sufficient performance capabilities and availability for evaluation in a variety of purposes. Data extraction of desired information from the scientific literature is one example of an arduous, time-consuming task by humans that AI could manage more quickly and easily. However, any new use of AI needs to be conceptually defined, prompts developed, and outcomes validated through performance testing. This invited perspectives article describes the process and author experiences using AI for the first time in a literature extraction task to assess trends in the reported validation parameters from a subset of publications involving mass spectrometry. The main conclusion was the AI more efficiently achieved the same quality of results as human experts in the field, but experience and care is needed to define and refine prompts to yield acceptable accuracy.