Comparison of machine learning algorithms for predicting admission in intensive care units based on the precision nursing framework

Background: Between 5% and 25% of patients with coronavirus disease 2019 (COVID-19) disease will require hospitalization in an intensive care unit (ICU) bed. The study aimed comparison of machine learning models for predicting admission of patients infected with corona virus disease in ICUs, based on the theoretical framework of Precision Nursing.<br/><br/>Methods: A retrospective, multicenter study, which included 547 patients positive for COVID-19 disease, admitted to hospitals. From the combination Precision Nursing variables, eight prediction models were tested: neural network, AdaBoost, logistic regression, random forest, K Nearest Neighbor, Naive Bayes, Support Vector Machine, decision tree. Evaluated by cross-validation, observing the metrics of area under the curve, sensitivity, specificity, and confusion matrix.<br/><br/>Results: The most important predictor variables were admission to ICU were age, admission to the university hospital linked to the Universidade Federal do Amazonas and the number of people living in the household. The decision tree model showed the best performance for area under the curve (0.668), sensitivity (0.633) and specificity (0.669).<br/><br/>Conclusions: The use of the model as a tool to support nurses’ decision making helps to identify downgrading of the condition, indicating early the patients most likely to be admitted to the ICU. A frequent evaluation of their predictive capacity are necessary, and it is believed that the inclusion of the vaccination variable could improve the predictive model performance.