Authors: Martino Recchia, Umberto Cornelli, Giovanni Belcaro.
Background: Model of the COVID-19 development are needed to have indication about its growth and decline over time.
Objective: To generate a model in relation to the Italian situation in the period between March 1st and April 22nd.
Methods: Five non-linear mathematical equations (Logistic 4P Rodbard; Weibull Growth; Logistic 4P; Logistic 3P; Exponential 3P) were used to describe the growth curves models relating to positive cases, deaths and their respective acceleration patterns. The analysed period is from 3rd March to 21st April.
Each model has been adapted separately to the observations using JMP14 software of SAS Institute. Next, all models were tested for goodness of Fit (quality of prediction) using Akaike’s information criterion (AICc), AICc Weight and Bayesian information criterion (BIC). The data were released by the Italian Heath Authorities and consistent for each of the 20 Italian Regions.
Results: The Rodbard 4PL Logistics equation was the best model for describing the trend of positive, death and differential cases, because it presented the lowest AICc, and BIC values and the highest AICc weight, compared tothe other models.
Conclusion: The data show that the COVID-19 reached the pick in the month of March, and in April was starting the decay of both positive cases and deaths. A large difference was shown among the Regions both in terms of number of cases and deaths. The shape of the curves show that the decay seems to be asymptotic.
View/Download pdf