A new framework testing the reliability of AI has been designed to address the global threat of antimicrobial resistance.
Image Credit: Scientific Frontline
Scientific Frontline: Extended "At a Glance" Summary: Explainable AI in Antibiotic Discovery
The Core Concept: A newly developed evaluative framework that tests the reliability, transparency, and chemical reasoning of artificial intelligence (AI) models used in the development of new antibiotics.
Key Distinction/Mechanism: Rather than accepting the "black box" nature of standard AI algorithms—which output predictions without explanation—this framework explicitly assesses an AI model's ability to interpret "activity cliffs," which are scenarios where minor chemical alterations drastically change a drug's effectiveness.
Major Frameworks/Components:
- Development and utilization of three distinct AI models trained on chemical compound datasets.
- Evaluation of AI efficacy using chemical compounds previously tested against the multidrug-resistant bacterium Staphylococcus aureus.
- Validation of the AI's ability to not only identify known antibiotic structures but also accurately explain what makes specific molecules active or inactive.
Branch of Science: Cheminformatics, Computational Biology, Pharmacology, Microbiology, Artificial Intelligence.
Future Application: This tool will integrate explainable AI directly into medical chemistry workflows, allowing researchers to make informed, resource-efficient decisions during the accelerated discovery of novel antibiotics.
Why It Matters: Antimicrobial resistance is a critical global health threat. By ensuring scientists can verify the chemical rationale behind AI predictions, this framework prevents misleading data from wasting laboratory resources and speeds up the deployment of life-saving treatments against superbugs.
The development of new antibiotics could be accelerated thanks to a new tool that tests the reliability of AI.
Researchers at the University of Queensland designed a framework to address the global threat of antimicrobial resistance, testing whether AI can provide reliable reasoning during antibiotic development.
Dr. Abdulmujeeb Onawole, from UQ’s Center for Superbug Solutions at the Institute for Molecular Bioscience, said drug-resistant bacteria are one of the greatest threats to global health, and there is an urgent need for new antibiotics.
“AI is revolutionizing drug development, but scientists struggle to trust its recommendations because the technology often cannot explain its reasoning,” Dr. Onawole said.
“We call this the ‘black box’ of AI—where AI provides an answer but there is no explanation of how it got there—and this prevents scientists from understanding the chemical reasoning behind its predictions.
“This lack of transparency is dangerous during antibiotic development, as misleading AI explanations can lead to incorrect decisions and wasted resources in the lab.”
Antimicrobial resistance, including resistance to antibiotics, is threatening healthcare globally by limiting effective treatment options against multidrug-resistant pathogens known as "superbugs."
“This is a high-stakes field, and while AI can help us save lives faster, we want to ensure the humans involved can make an informed decision,” Dr. Onawole said.
“Longer term, this could contribute to the faster discovery of new antibiotics to combat drug-resistant superbugs.”
In the study, researchers developed three AI models using datasets of chemical compounds previously evaluated against the superbug bacterium Staphylococcus aureus.
The framework was tested on each AI model and examined whether AI could correctly identify important drug structures and interpret "activity cliffs"—scenarios where small chemical changes altered a drug’s effectiveness.
Dr. Johannes Zuegg, from UQ’s Center for Superbug Solutions, said the research found all three AI models were good at spotting known antibiotic structures but differed significantly in their ability to explain what made a molecule active or inactive in developing antibiotics.
“We have shown our framework can successfully assess whether AI systems can provide trustworthy chemical explanations, which is critical to medical chemists in drug development,” he said.
“This is an important step toward speeding up the integration of AI into antibiotic research.”
Published in journal: Journal of Cheminformatics
Title: Framework for evaluating explainable AI in antimicrobial drug discovery
Authors: Abdulmujeeb T. Onawole, Mark A. T. Blaskovich, and Johannes Zuegg
Source/Credit: University of Queensland
Edited by: Scientific Frontline
Reference Number: cobi062626_01
.jpg)