Sudden and/or arrhythmic death (SAD), which typically results from lethal ventricular arrhythmias (ventricular tachycardia and ventricular fibrillation, VT/VF) in the setting of coronary heart disease (CHD), afflicts an estimated 310,000 persons annually in the United States. Reductions in SAD have continued to lag those observed for other coronary heart disease (CHD) outcomes despite advances in resuscitation therapies and the use of implantable cardioverter-defibrillators (ICDs). Current approaches to SAD prevention remain centered on placing ICDs in patients with left ventricular ejection fraction (LVEF) <30-35% – even though the majority of SAD occurs in the setting of LVEF >30-35%. In effect, the proportionately larger segment of the at-risk population has been understudied and thus undertreated. Despite this unmet need, there remain very few, if any, prospective studies examining SAD risk prediction in individuals with CHD and LVEF >30-35% over a long enough time horizon where ICD therapy might be cost-effective. For this very reason, the PRE- DETERMINE Cohort Study was intentionally designed to address this scientific gap and prospectively study clinically relevant approaches to SAD risk prediction in CHD patients with LVEF >30-35%. In this application, we propose to leverage the originally NHLBI-funded base cohort resource to continue adjudication of accruing SAD and VT/VF events, in addition to competing causes of death, to attain 10+ years of endpoint adjudication to enable the development and validation of multi-marker SAD risk prediction models based on combinations of multi-dimensional clinical, ECG, imaging, biomarker, and genetic data generated in this unique multicenter cohort of 5761 CHD patients. We will also leverage the base cohort to interrogate novel fatty acid derived eicosanoids and putative arrhythmia modulating proteomic analytes in relation to risk for SAD and competing causes of mortality in patients with CHD. Novel methods of competing risk analyses will be used to integrate absolute and proportional SAD risk into SAD risk prediction models and to elucidate separate associations between SAD vs. non-SAD causes of death. Machine learning approaches will be applied to uncover inter-relations and latent features from multi-modality data not easily detected by conventional models. An overarching goal of our work is to accurately identify those individual subsets of the broader population who have sufficiently high absolute and proportional risk for SAD that they warrant inclusion in randomized trials of primary prevention ICD therapy. The aims of the current proposal also offer new opportunities to identify potential mechanistic pathways underlying the genesis of lethal ventricular arrhythmias that could serve as novel targets for SAD prevention – extending beyond ICD placement – in patients with CHD and possibly even in the general population wherein CHD underlies most SAD events. The continuation and expansion of the PRE...