Discovery

UC Irvine takes AI-powered metabolic imaging into drug resistance testing

UC Irvine takes AI-powered metabolic imaging into drug resistance testing

University of California, Irvine has received a USD 3 million RC2 grant from the National Center for Advancing Translational Sciences (NCATS) to develop a fluorescence lifetime imaging microscopy (FLIM) and machine learning platform designed to improve how drugs are tested against live cells under near-physiological conditions.

The award addresses a recognized gap in preclinical drug evaluation: standard laboratory assays are conducted in conditions that poorly replicate human physiology. The platform, termed FLIM-based drug testing (FLIM-DT), captures real-time, label-free metabolic signatures of bacteria and other cells within minutes of drug exposure — a speed advantage over conventional culture-based methods. Preliminary data cited in the grant indicate FLIM can detect metabolic changes in bacteria shortly after antibiotic exposure, a finding the team will use as the basis for antimicrobial susceptibility testing (AST) as the prototype application.

The machine learning component is intended to classify multidimensional metabolic imaging data rapidly enough to support clinical decision-making. A key translational claim in the application is that FDA-approved antibiotics classified as ineffective by standard AST can show activity against multidrug-resistant pathogens when tested under physiological media conditions — a finding the team said it will systematically evaluate using a diverse bacterial pathogen and antibiotic dataset. The platform is also described as adaptable to cancer and neurodegenerative disease drug testing, though the funded work focuses on infectious disease as the initial use case.

The AllSci BriefSystematic R&D and deal news. Daily.

Peter Chang leads the project alongside co-investigators Erlinda Rose Ulloa and Michelle Digman. The interdisciplinary team spans microbiology, advanced imaging, and machine learning. The grant supports a phased implementation strategy that positions FLIM-DT initially as a reflex or add-on test for drug-resistant infections within the UC Irvine Clinical and Translational Science Award (CTSA) hub network.

Existing automated AST platforms operate on similar turnaround timescales but rely on growth-based endpoints in standard media. The FLIM-DT approach differs by reading metabolic state rather than replication, and by using physiologically modeled media — a distinction the team argues will improve predictive accuracy for clinical outcomes. No head-to-head comparative data with commercial AST systems are disclosed in the current award.


Spot something wrong? Report an issue with this article