Project Summary Introduction: Jason Coult, PhD, is a scientist whose career goal is to become an independent computational investigator improving survival from out-of-hospital sudden cardiac arrest (OHCA) through research of novel OHCA resuscitation technologies that incorporate clinical understanding. His K01 proposal, “Artificial Intelligence to Improve Resuscitation following Out-of-Hospital Cardiac Arrest”, seeks to develop smarter, deep learning-based defibrillator algorithms that can improve OHCA survival by guiding personalized resuscitation treatment. Candidate: Dr. Coult is a recently- appointed Research Assistant Professor at the University of Washington Department of Medicine. He completed a PhD in bioengineering in 2019, and has expertise in signal processing, artificial intelligence (AI), and OHCA research. Career development and mentorship: Dr. Coult has convened an interdisciplinary team of mentors and advisors with expertise in animal and human clinical resuscitation, cardiac electrophysiology, prehospital emergency medical services (EMS), AI and deep learning, computational arrhythmia models, biostatistics, and defibrillators. The training plan cultivates necessary skills in deep learning and biostatistics, advances understanding of clinical resuscitation, provides career guidance, and facilitates Dr. Coult’s progression to become a successful independent investigator. The proposed work will take place under Thomas Rea MD (primary mentor) at the University of Washington. Proposed research: OHCA is a leading cause of mortality. Survival is possible though generally poor. Resuscitation currently follows a fixed, one-size-fits-all protocol that requires CPR interruption at regular intervals to determine the patient’s cardiac rhythm, assess vital status, and apply treatment. This research will use large retrospective datasets of human OHCA defibrillator recordings, Dr. Coult’s expertise and mentorship/advisory group, and emerging deep learning methods to achieve the following aims: 1) to design deep learning-based AI algorithms that can identify the specific OHCA rhythm (asystole, ventricular fibrillation (VF), organized rhythm) and estimate the “vitality” (morphologic phenotype associated with clinical outcome) of these rhythms during ongoing CPR, 2) to characterize the effect of drug interventions on measures of vitality, and 3) to apply rhythm and vitality features to predict shock-refractory VF patients (requiring ≥ 3 shocks) during CPR, enabling preemptive antiarrhythmics or other strategies to mitigate subsequent shock failure. Summary: The proposed research will advance resuscitation towards a precision strategy that aligns treatment with an individual patient’s real-time physiology, providing the potential to improve OHCA survival. The training and mentorship will foster the development of necessary AI-related technical skills, deepen clinical understanding of OHCA and resuscitation, and help inform next- step, R01-funded res...