Forecasting the cumulative number of COVID-19 deaths in China: a Boltzmann function-based modeling study

Gao, Yuanyuan; Zhang, Zuqin; Yao, Wei; Ying, Qi; Long, Cheng; Fu, Xinmiao

Collectively, all sets of data from both COVID-19 deaths and the 2003 SARS deaths were well fitted to the Boltzmann function. We propose that the Boltzmann function is suitable for analyzing not only the cumulative number of confirmed COVID-19 cases, as we recently reported 3 (Fig. 1A) , but also numbers of deaths, as reported here. Modeling studies on the COVID-19 outbreak have been performed, 6 and COVID-19 deaths have been estimated by other groups using different models. For example, using a datadriven analysis, Li et al 7 recently predicted that total deaths in Hubei would be 2,250, much lower than the observed number, 2,761 (as of February 29). Using the susceptible-infectedrecovered-dead model, Anastassopoulou et al 8 predicted that total deaths might exceed 7,000 by February 29, but this number was much higher than the actual figure reported.