Project Summary/Abstract Rapid identification and antimicrobial susceptibility testing (AST) of clinical isolates are crucial to ensure early appropriate treatment and prevent misuse of broad-spectrum antimicrobials. Both phenotypic (i.e., growth, inhibition, or killing event-based) and molecular (i.e., resistant gene or molecule-based) ASTs are routinely used in the hospital clinical microbiology laboratory. The limitations of molecular methods (e.g., qPCR, MALDI-TOF MS) include their narrow scope (i.e., select resistance genes/molecules), or in some cases the inability to predict antimicrobial susceptibility, while phenotypic methods involve time-consuming, multi-step expansion culture resulting in prolonged time to result, which can delay appropriate treatment decisions. This proposal leverages an exclusive liquid repellency (ELR) based under-oil open microfluidic system for a transformative approach to next-generation phenotypic ASTs. With a multidisciplinary team including clinicians, microbiologists/pharmacologists, surface scientists, and bioengineers, we outline three aims for the development of this next-generation, phenotypic AST system. The ELR-based AST system aims to meet the following criteria: i) direct AST using the original clinical isolates (e.g., from blood, sputum, urine, abscess) to eliminate time-consuming expansion culture and passage- associated selection bias ex vivo, ii) comprehensive test coverage including anaerobes, multispecies communities, heteroresistance (i.e., resistant mutants within the wild-type population), and iii) rapid AST with the goal of sample to report in less than 4 hours. In Aim 1 we propose to develop ELR-centrifugation and small-volume ( μ l scale) lossless sample processing for isolation, enrichment, and preparation of sparse (i.e., 1-100 cfu/ml) bacteria from whole blood. The goal is to directly isolate and enrich bacteria from whole blood without expansion culture. We will combine the lossless sample processing enabled by ELR with lysis-centrifugation to maximize the recovery yield of bacteria (> 90%) from whole blood. In Aim 2 we will demonstrate compatibility of ELR AST with intrinsic fluorescence label free imaging modalities and deep learning-based identification. The goal is to develop and apply proof-of- concept advanced label-free, single-cell resolution, live-cell imaging, and deep learning algorithms to integrate bacterial detection, species identification, and antimicrobial screening thereby eliminating isolate passage expansion. In Aim 3 we propose to develop label-free, direct detection of heteroresistance in priority human pathogens in clinical isolates. The goal is to detect heteroresistance [i.e., a small/rare subpopulation (e.g., <1%) of resistant mutants in clinical isolates], which cannot be identified by standard clinical methods. We will utilize sweep distribution to array small numbers of bacteria directly from clinical isolates, thus enabling the detection of these heterore...