Abstract 69: Predictors Of Cardiac Arrest-related Hospitalizations In Young (18-44 Years) Females - An Artificial Neural Network Analysis Using A Nationwide Cohort

Abstract

Background: Considering the limited availability of data on Cardiac Arrest (CA) in young patients and especially females, we aimed to determine the predictors of CA in this population using Artificial Neural Network (ANN) Model in a national cohort from the United States. Methods: We identified CA-related hospitalizations among young females (18-44 years) using 2018’s National Inpatient Sample database. ANN’s predictive factors were selected for this cohort. Young females with CA (n=10810, 0.2% of all 2018 young female admissions) were randomly split into training data (n=7567, 70%) which were used to calibrate ANN and testing data (n=3243, 30%) which were used to evaluate the accuracy of the algorithm. We compared the frequency of incorrect prediction between training and testing data and measured the Area under Receiver Operator Curve (AUC) to determine ANN’s efficacy in predicting CA. Results: Young females with CA often consisted of older (median age 36 vs 30 years), blacks (25.3% vs 18%), and patients from lower-income quartile (0-25% income quartile:36.4% vs 29.9%) with higher rates of modifiable cardiovascular disease risk factors vs. females admitted without CA (p<0.001). Females with CA expectedly had significantly high (48.4%) in-hospital mortality. Normalized Predictors are displayed in Table 1. Our ANN model had AUC 0.902 (Fig 1) which correlates with an excellent prediction model. Our data showed 0.2% error in both testing and training models. Conclusion: Our ANN model achieved high performance to predict risk factors for CA admissions in young females. It will enable clinicians to screen high-risk young female hospitalized patients and improve survival in them.