AUTHOR=Robledo Almonacid Juan Eduardo , Lombardo Christian , Romano Mariana , Quiroga Agustina , Cambuli Bianchi Paula , Hualpa Mauricio , Giai Constanza , Oviedo Xiomara María A. , Salgado Mansur Ramiro Alejo , Vallejo Mariana Guadalupe , Quintero Cristián Andrés 

TITLE=The WHO’s critical bacteria list: scientific response eight years after its implementation and development of an AI-based tool for its monitoring

JOURNAL=Frontiers in Pharmacology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2025.1633382

DOI=10.3389/fphar.2025.1633382

ISSN=1663-9812

ABSTRACT=BackgroundIn 2017, the World Health Organization (WHO) issued a global alert identifying 12 bacteria in urgent need of new treatments.Main bodyThis study assesses the scientific community’s response to this alert by analyzing original research publications using LLMzCor, an AI-based tool developed and validated by our group. To compare trends, we focused on publications from 5 years before and after the alert, specifically on three bacteria listed in the WHO alert, sorted by priority level: Acinetobacter baumannii (Critical), Shigella spp (High), and Neisseria gonorrhoeae (Medium) and three non-listed as controls (Rickettsia spp., C. trachomatis, and C. difficile). Articles were classified into three categories: (i) identification of Resistant strains, (ii) development of New treatments, and (iii) Immunization strategies.ResultsAlthough overall publications increased after the WHO alert, no statistically significant changes were found in the reports of Resistant strains over time. The development of New treatments for the listed bacteria showed a slight increase, between 2% and 10%. Furthermore, Immunization strategies remained relatively unchanged, with less than 2%. Meanwhile, LLMzCor demonstrated robust performance across categories, F1-scores ranging from 0.65 to 0.72 in key classifications, while recall peaked at 0.75, indicating a high capacity to identify relevant articles. These results support the model’s reliability for large-scale automated classification of scientific abstracts.ConclusionThese findings, supported by LLMzCor, underscore the urgency of a stronger WHO alert and action plans to develop new strategies against bacterial resistance.