Home Business Bio Tech AI and Explainability Could Re...

AI and Explainability Could Revolutionize Antibiotic Discovery

Bio Tech

Explainable AI, XAI, antibiotic discovery, AI models, drug development

Business Honor
21 August, 2024

New research uses Explainable AI (XAI) to improve drug models, potentially leading to better antibiotics and addressing resistance challenges.

Artificial Intelligence (AI) is transforming various fields, from self-driving cars to automated email corrections. Now, AI's potential extends into drug discovery, particularly for developing new antibiotics. Researchers are harnessing Explainable AI (XAI) to gain insights into AI decision-making processes, potentially leading to breakthroughs in antibiotic development.

At the upcoming American Chemical Society (ACS) fall meeting, researchers from the University of Manitoba will present their innovative use of XAI to inspect predictive AI models. Unlike traditional AI models, which often operate as "black boxes," XAI aims to provide transparency by revealing how decisions are made, which is crucial for fields like chemistry where understanding the rationale behind predictions is vital.

Rebecca Davis, a chemistry professor at the University of Manitoba, and her team have employed XAI to analyze AI models predicting the biological activity of drug molecules. By integrating XAI with their AI models, they can pinpoint which molecular features are deemed important by the AI for antibiotic activity. For instance, XAI revealed that, contrary to conventional belief, the penicillin core's attached structures, rather than the core itself, were crucial for its classification.

This approach not only enhances understanding of AI's decision-making but also refines the training of predictive models. The researchers are now collaborating with microbiologists to synthesize and test compounds predicted by their improved AI models. Their goal is to develop new antibiotic candidates that could outpace resistant pathogens.

The project, funded by the University of Manitoba, the Canadian Institutes of Health Research, and the Digital Research Alliance of Canada, demonstrates how XAI can make AI's role in drug discovery more transparent and trusted. This advancement could lead to more effective antibiotics and a better understanding of AI in chemistry.