Computational Intelligence Applied to Digital Health as a Tool for Analysis and Support for Decision Making: the regulation of covid-19 beds and vascular surgeries in the state of Rio Grande do Norte
covid-19, machine learning, deep learning, bed regulation, RegulaRN.
The covid-19 was one of the biggest pandemic outbreaks in recent years. Due to the rapid spread, the impacts of the disease affected several countries around the world, putting pressure on government institutions to develop devices to increase the number of beds and enable monitoring. In Rio Grande do Norte, a Brazilian state, RegulaRN was the system used to regulate the beds of patients with covid-19 for public hospitals and contracted with the state. In this article, we explore the use of machine learning and deep learning techniques on RegulaRN data in order to identify the best models and parameters capable of predicting the outcome of an inpatient. We analyzed the RegulaRN database, about 25,366 regulations, from April 2020 to August 2022. We selected the nine most pertinent features out of twenty possible ones, removed blank or inconclusive data, preprocessed, trained and tested. The results point to better performance in evaluating the metrics of the multilayer perceptron with the adam optimizer, with around 84% accuracy. Finally, we discuss the main impacts of using the model to help bed regulators during the regulation process, as well as aspects of public health resource management and how these models can help government actions in health crises.