ABSTRACT
The response to the ongoing second wave of the COVID-19 pandemic can be helped by giving medical professionals access to models learned on patient data. To achieve this, we learned a Bayesian network model to predict risk of ICU admission, death and time of stay in the hospital from patient history, initial vital signs, initial laboratory tests and medication. Data were obtained from patients that were admitted to an HM hospital with suspicion of COVID-19 until 24/04/2020, excluding unconfirmed diagnosis, those who were admitted before the epidemic started in Madrid, had an outcome that was not discharge or death or died within 24 hours of presentation. Relevant variables for the model were selected with help from medical professionals. We learned the model using Bayesian search as implemented in GeNIe. Of 2,307 patients in the dataset, 679 were excluded. With the remaining 1,645 patients, we learned a model that predicted death with 86.4% accuracy. Some of the initial variables were discarded because they were independent of the outcomes of interest conditioned on some of the other variables. This high redundancy might be useful to build simpler tests for the severity of COVID-19. We show how the model can be used at different stages of patient admission and even with only partial information about the patient. This can be done by clinicians that want a fast second opinion or a summary of the available data from previous patients similar to the one at hand. We then include how we plan to improve the model with extra patient data and how it could be expanded to other contexts, like for example, an epidemiological one.
ABSTRACT
ObjectivesTo present a model that enhances the accuracy of clinicians when presented with a possibly critical Covid-19 patient. MethodsA retrospective study was performed with information of 5,745 SARS-CoV2 infected patients admitted to the Emergency room of 4 public Hospitals in Madrid belonging to Quiron Salud Health Group (QS) from March 2020 to February 2021. Demographics, clinical variables on admission, laboratory markers and therapeutic interventions were extracted from Electronic Clinical Records. Traits related to mortality were found through difference in means testing and through feature selection by learning multiple classification trees with random initialization and selecting the ones that were used the most. We validated the model through cross-validation and tested generalization with an external dataset from 4 hospitals belonging to Sanitas Hospitals Health Group. The usefulness of two different models in real cases was tested by measuring the effect of exposure to the model decision on the accuracy of medical professionals. ResultsOf the 5,745 admitted patients, 1,173 died. Of the 110 variables in the dataset, 34 were found to be related with our definition of criticality (death in <72 hours) or all-cause mortality. The models had an accuracy of 85% and a sensitivity of 50% averaged through 5-fold cross validation. Similar results were found when validating with data from the 4 hospitals from Sanitas. The models were found to have 11% better accuracy than doctors at classifying critical cases and improved accuracy of doctors by 12% for non-critical patients, reducing the cost of mistakes made by 17%.
