Analysis of COVID-19 infection waves in Japan by Avrami equation

TL;DR

This paper models the multiple COVID-19 infection waves in Japan from 2020 to 2021 using the Avrami equation, achieving high accuracy in simulation and enabling future infection prediction.

Contribution

It extends previous work by simulating all infection waves with superposition of five waves and applies least-squares fitting to determine parameters.

Findings

Simulation accuracy exceeded 95% confidence interval after the second wave

The model effectively predicts near-future infection numbers

Wave structure analysis helps understand infection dynamics

Abstract

Last time, an attempt was made to analyze the first wave of COVID-19 in Japan from February to May 2020 by the Avrami equation. This time, all infection waves that occurred repeatedly from February 2020 to September 2021 were simulated by developing the last work. The entire waveform was basically simulated by the superposition of five major waves, and the least-squares method was applied to each wave to determine the parameters of the equation. Despite the simulation using the single equation assuming that the parameters were constants, the accuracy of the simulated cumulative infection D was as good as 95%CI / D < 2.5% after the second wave when D exceeded 10,000. Since the simulation was highly accurate, it was effective in predicting the number of infections in the near future and in considering the cause of changes in the number of infections observed as the detailed structure of the waveform.