Dascena publishes on mortality prediction in alcohol use disorder inpatients
In this study Dascena validates a mortality prediction algorithm in a unique patient population. The findings are published in Computers in Biology and Medicine.
Chronic alcohol use can confound the diagnosis of a patient, leading to a decrease in the predictive value of existing systems designed to analyze patient stability. AutoTriage is an algorithm that uses data available in the electronic health record (EHR) to predict patient mortality. Through multi-dimensional analysis, AutoTriage is able to generate accurate mortality risk scores, featuring an 80% specificity at 90% sensitivity and an area under the receiver operating characteristic curve (AUROC) of 0.93 for 12-hour mortality prediction in the alcohol use disorder (AUD) patient population. With 81% accuracy and an odds ratio of 36, AutoTriage outperforms existing mortality prediction tools that include Modified Early Warning Score (MEWS), Sepsis-Related Organ Failure Assessment (SOFA), and Simplified Acute Physiology Score (SAPS II). These results suggest that AutoTriage can be an effective tool for accurate patient stability predictions in AUD patients.