Forecasting the Number of Patients with COVID-19 in the Lviv Region
Keywords
Abstract
The dynamic of new cases of COVID-19 infections in Lviv district was investigated. With this purpose, the statistic data from the official site of COVID-19 distribution monitoring on Ukraine was collected. These data contains daily statistics on hospitalized persons with suspected and confirmed cases of the disease and statistics on recovered and deaths in Ukraine. In the paper the dependency between the grouch of the patients amount and the reduce of quarantine restrictions was determined.
The existing publications on the COVID-19 spread forecast in Ukraine were reviewed. In these works, authors were using methods of the intelligent data analysis, artificial neural networks, exponential forecast, similarities, correlation and regressive analysis. The exclusive attention was paid to the use of Back Propagation Neural Networks for the short-term forecast of the amount growth of COVID-19 patients in Ukraine. The methods of technical analysis of the time serials based on the use of basic indicators like “zigzag” and “supertrend” for the patients amount forecast in Lviv district were used as well.
The non-iterative neural network of the radial basis functions with additional inner-layer connections between the hidden-layer neurons was applied to the forecast of confirmed cases of COVID-infections in Lviv district. As a short-term forecast was built, considering predictions for one day. As a middle-term forecast, predictions for two weeks were done and also the method of the “sliding window” was used. The same approach was used to make a 1-day and two weeks forecast of the amount of patients recovering and deaths cases for the Lviv district.
Based on these forecasts the methodology to control the introduced quarantine restrictions in Lviv district was offered. Taking into account the middle-forecast results, there will be no recommendations to do any next stage quarantine restrictions reduce in May 29th. In addition, the required amount of beds that have to be provided at this date in base-hospitals was calculated.
