PROJECT SUMMARY/ABSTRACT For epidemiological studies, future clinical trials, and personalized patient care, there is a critical need to create a risk-stratification system for microbial keratitis. Microbial keratitis (MK), a debilitating, infectious corneal disease, is estimated to be the fourth-leading cause of blindness worldwide. MK severity depends on a complex interaction of patient, organism, and environment, resulting in a spectrum of clinical presentations and responses to treatment. Clinical presentations manifest with unique morphology features and clinical symptoms. Morphology features are visible in the cornea, and symptoms are measurable. But most patients are treated with non-specific broad-spectrum antimicrobials, an approach that increases antimicrobial resistance. This non-specific treatment approach lacks congruence with the unique MK presentations. There is a critical need for a new strategy to personalize treatments for MK and measure treatment efficacy. With quantified MK morphologic and clinical features, clinicians will have the tools to risk-stratify patients. The long-term goal is to develop rapid, objective, personalized treatment plans for patients with MK. This proposal’s objective is to quantify dynamic morphologic and clinical MK features using image and electronic health record (EHR) analyses and then build a risk-stratification scoring system associated with MK outcomes. The proposed research will test the hypothesis that morphologic and clinical features accurately risk- stratify patients for corneal and vision outcomes. Our premise is supported by preliminary data demonstrating that: (1) different organisms generate distinct morphologic and clinical features; (2) clinicians quantify morphology less precisely than image-analysis methods, (3) an expert is able to use MK features to tailor treatments; (4) the use of quantified features has improved outcomes in other diseases, such as diabetic retinopathy, by helping providers to tailor treatments; (5) EHR data can be used to quantify and classify clinical disease features accurately; and (6) EHR data can be used effectively to risk-stratify patients. Aim 1 will develop objective image analysis tools to measure features of MK with existing clinical equipment. Aim 2 will evaluate MK treatment efficacy using morphologic image analysis and clinical features from prospective surveys. Aim 3 will risk-stratify patients with MK by combining image analysis and EHR extracted data. The expected outcomes are: (1) characterized databases of MK images and linked clinical data, (2) quantified MK features across a spectrum of clinical presentations, (3) performance-tested, open-source imaging algorithms and surveys to measure MK markers dynamically, and (4) a novel risk stratification model and scoring system. The resultant work will have significant value to clinicians. Clinicians can use practical, low-cost technologies and readily-available EHR data to quantify MK features and ...