Neural network analysis of clinical variables predicts escalated care in COVID-19 patients: a retrospective study

Abstract

This study sought to identify the most important clinical variables that can be used to determine which COVID-19 patients hospitalized in the general floor will need escalated care early on using neural networks (NNs). Analysis was performed on hospitalized COVID-19 patients between 7 February 2020 and 4 May 2020 in Stony Brook Hospital. Demographics, comorbidities, laboratory tests, vital signs and blood gases were collected. We compared those data obtained at the time in emergency department and the time of intensive care unit (ICU) upgrade of: (i) COVID-19 patients admitted to the general floor (N = 1203) vs. those directly admitted to ICU (N = 104), and (ii) patients not upgraded to ICU (N = 979) vs. those upgraded to the ICU (N = 224) from the general floor. A NN algorithm was used to predict ICU admission, with 80% training and 20% testing. Prediction performance used area under the curve (AUC) of the receiver operating characteristic analysis (ROC). We found that C-reactive protein, lactate dehydrogenase, creatinine, white-blood cell count, D-dimer and lymphocyte count showed temporal divergence between COVID-19 patients hospitalized in the general floor that were upgraded to ICU compared to those that were not. The NN predictive model essentially ranked the same laboratory variables to be important predictors of needing ICU care. The AUC for predicting ICU admission was 0.782 ± 0.013 for the test dataset. Adding vital sign and blood-gas data improved AUC (0.822 ± 0.018). This work could help frontline physicians to anticipate downstream ICU need to more effectively allocate healthcare resources.

Keywords

• Machine learning
• Coronavirus
• Pneumonia
• SARS-CoV-2
• Predictive model