Abstract 11020: Predictors of Cardiac Arrest Including Ventricular Fibrillation in Young Patients Hospitalized With Hypertrophic Obstructive Cardiomyopathy: An Artificial Neural Network Analysis Using a Nationwide Cohort

Abstract

Background: It is crucial for providers to screen for high-risk patients with limited data being available on young hypertrophic obstructive cardiomyopathy (YHOCM) patients. We aimed to determine predictors of cardiac arrest including ventricular Fibrillations (VF) in this population using the Artificial Neural Network (ANN) model in a national cohort. Methods: We identified young patients (18-44 years) with HOCM from National Inpatient Sample (2018) with relevant ICD 10 codes. ANN’s predictive factors were selected for cardiac arrest including VF. YHOCM were randomly split between training (70.8%) and testing (29.2%) datasets. Training data was used to calibrate ANN whereas testing data was used to evaluate the accuracy of the algorithm. We compared the frequency of incorrect prediction between training and testing data and measured area under the Receiver Operating Curve (AUC) to determine ANN’s efficacy in predicting cardiac arrest. Results: YHOCM (2280, median age 36 years, 52.4% males, 47.6% females, 50.9% Whites, 32.2% Blacks, 11.6% Hispanics) had all-cause mortality of 21.1% with compared to that of 1.1% without cardiac arrest. Training data showed a lower rate of 3.9% vs 4.7% of error in predictions vs testing, thereby depicting better accuracy. Normalized predictors are displayed in Fig 1a. The AUC was 0.799 (Fig. 1b) which shows an excellent ANN model for cardiac arrest in YHOCM hospitalizations. Conclusions: Our ANN model successfully revealed predictors of cardiac arrest including VF in YHOCM patients that can be utilized to improve/prevent such events in high-risk patients leading to a better quality of life and survival rates.