A Novel Machine Learning-Enabled Metabolic Imaging Platform for Physiologically Relevant, Predictive Drug Efficacy Testing - PROJECT SUMMARY/ABSTRACT The failure of traditional laboratory assays to mimic human physiology remains a critical bottleneck in drug development, resulting in poor predictions of clinical efficacy and leaving patients with aggressive cancers, neurodegenerative diseases, and drug-resistant infections without effective therapies. We propose a transformative platform integrating fluorescence lifetime imaging microscopy (FLIM) with machine learning (ML) to deliver real-time, label-free metabolic profiling of live cells under near-physiological conditions. By capturing nuanced drug responses missed by conventional assays, our FLIM-based drug testing (FLIM-DT) platform promises to dramatically improve the accuracy, speed, and scalability of drug efficacy evaluation. As a prototype, we will deploy this technology for antimicrobial susceptibility testing (AST), using physiologically modeled media and a diverse dataset of bacterial pathogens and antibiotics to establish clinical utility and scalability. This project addresses a critical translational gap by building on foundational work showing that FDA-approved antibiotics deemed “ineffective” by standard testing can exhibit potent therapeutic benefits against multidrug-resistant pathogens under physiological conditions. Our innovative integration of advanced metabolic imaging with AI analytics represents a novel approach broadly adaptable beyond infectious diseases, including cancer and neurodegenerative disorders. Leveraging the strengths of our CTSA hub, we will engage clinicians, microbiologists, and health system leaders to ensure real-world feasibility, sustainability, and rapid clinical adoption. Our phased implementation strategy will focus initial planning on FLIM-DT as a reflex or add-on test for drug-resistant infections—maximizing future clinical impact while minimizing workflow disruption. By enabling rapid and accurate drug efficacy assessments, our platform can significantly shorten diagnostic timelines and improve patient-specific treatment strategies, reducing morbidity and healthcare costs. Through this approach, we will generate feasibility data, refine workflows, and develop a robust roadmap for broader dissemination. Ultimately, this project lays the foundation for a new era of personalized, predictive drug testing with broad applications across medicine. Our interdisciplinary team, combining expertise in microbiology, imaging, and machine learning, will pursue three focused aims: adapting FLIM-DT for physiological drug testing, integrating ML for rapid analysis, and evaluating clinical feasibility via our CTSA hub. Preliminary data demonstrate FLIM’s ability to detect metabolic changes in bacteria within minutes of antibiotic exposure, highlighting its potential to revolutionize antimicrobial susceptibility testing and beyond. By bridging the translational divide between discovery and patient care, our platform directly advances NCATS’ mission to accelerate the delivery of effective, individualized therapies and improve public health outcomes. The knowledge gained promises immediate translational impact—overcoming drug evaluation roadblocks, enabling drug repurposing, and catalyzing a paradigm shift in drug discovery across medicine.
