Pathnostics developed a machine-learning model that predicts antibiotic susceptibility for urinary tract infections using PCR results and basic demographics, offering clinicians actionable guidance within 8 hours—well before full phenotypic results return. Presented at AMP 2025 by Jim Havrilla, PhD, the model draws on 32,300 paired PCR–phenotype samples and uses light gradient boosting with extensive cross-validation to optimize calibration. The system includes 20 antibiotic-specific algorithms and reports only probabilities above 60% to avoid unreliable predictions. Early results show strong alignment with phenotypic susceptibility, especially for nitrofurantoin, fosfomycin, and beta-lactam combinations. A pilot launch is planned for 2026.