Intensive Care Unit Occupancy Time Series Forecasting for COVID-19 Pandemic

Nowadays, the pandemic situation because Covid-19 affected everything related to the public health infrastructure. One of the most important aspects to treat the Coronavirus contingency was the hospitalization of patients with severe cases, where some cities had high demand of beds in the intensive care units for attending these people. From the engineering tools, time series forecasting has been an alternative for analyzing data in this context. This work proposes to do a forecasting on the demanded beds for the severe cases of Covid-19, employing models from the artificial intelligence. Techniques as fuzzy and neural networks were tried to determine the necessary number of beds through the use of the registration of this information daily, in a Latin American city. Results show that strategies based on neural networks were better compared to fuzzy models, according to the root mean square error. The obtained models can contribute to the authorities in making decision process to react before the similar contingencies.