SBIR Phase II: Leveraging Sequencing to Identify and Predict Multidrug Resistance

Award ID: 2604976

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase II project is to develop a rapid diagnostic platform that predicts both current antibiotic resistance and the risk of treatment-emergent resistance within 24 hours, enabling physicians to select precise, effective treatments immediately. By enabling precision medicine in infectious disease treatment, this technology has the potential to improve patient outcomes, reduce healthcare costs associated with resistant infections, decrease unnecessary antibiotic use, and help preserve the effectiveness of existing antibiotics. The commercial impact extends to hospitals facing financial penalties for poor infection control and insurance companies seeking to reduce costs under value-based care models. This Small Business Innovation Research (SBIR) Phase II project develops a machine learning platform that analyzes bacterial whole genome sequencing data to predict antibiotic resistance through mutational signature analysis. Unlike traditional approaches that detect only known resistance genes, this platform examines patterns of mutations across the entire bacterial genome to identify signatures of past antibiotic exposure and genetic instability. The research objectives include expanding the platform from a single pathogen to cover multiple clinically important bacterial species, developing models to predict treatment-emergent resistance that can arise during therapy, and validating these predictions using clinical samples from multiple medical centers. The technical approach uses non-negative matrix factorization to extract mutational signatures from large datasets of bacterial genomes, then trains recurrent neural networks and other machine learning models to predict both current resistance phenotypes and future resistance development. Phase I feasibility studies demonstrated models achieving over 90 percent accuracy for several antibiotics and 100 percent accuracy in predicting treatment-emergent resistance in a small clinical cohort. The anticipated technical results include validated predictive models across multiple pathogen-antibiotic combinations and demonstration of clinical utility in prospective studies. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

NSF Program Director: Henry Ahn