Introduction: Drug-induced QT prolongation causes significant morbidity and mortality but could be preventable with prediction of susceptible individuals. Machine learning (ML) algorithms applied to electronic health record (EHR) data may provide a method for identifying these individuals and could be automated to directly alert providers of risk. This study applies ML techniques to EHR data to identify an integrated model that can be deployed to predict risk of drug-induced QT prolongation. Methods: We examined data from the UCHealth EHR that has been harmonized to the Observational Medical Outcomes Partnership common data model and identified inpatients who had received a medication known to prolong the QT interval and had an electrocardiogram (ECG) performed within 24 hours. We used a binary outcome of the development of a QTc interval > 500 ms by ECG within 24 hours of medication initiation or no ECG with a QTc interval > 500 ms throughout the encounter. We compared multiple machine learning methods (logistic regression, random forest, naïve Bayes, and deep neural network (DNN)) by classification accuracy as assessed by AUC and F 1 score. We performed calibration and scaling of the final model. Results: We identified 35639 inpatients who received a known QT-prolonging medication and had an ECG within 24 hours. Of those, 4558 patients developed a QTc > 500 ms and 31081 patients did not. The DNN model provided reasonable classification accuracy (F1 score 0.404; AUC 0.71) and was the most accurate method tested. A range of decision cutpoints were plotted with respective sensitivity/specificity (Figure 1). Conclusions: We found that by applying a DNN to EHR data, we could reasonably predict individuals susceptible to drug-induced QT prolongation. Varying cutpoints can be used to tailor the model to the desired sensitivity. Future studies are needed to validate this model in novel EHRs and within the physician order entry system to assess ability to improve patient safety.