Data source: JHU humandata

Model settings

General parameters: N: 80000000 e: 2 p: 7 q: 15

Time range for analysis: to . Time axis: .

Intervals A_i = [t_i+1, t_i+1] notated by t_i. Last number is total length. If you provide only one number, this number will be used as total length and we use the empirical values for a(k) directly. All intervals will be shifted by t₁=. If you want to extend your last interval for simulation, just add a slash followed by number after the last interval, e.g., enter 0, 100 / 120 for A₁=[t₁, t₁+100) and use the mean of these a(k) for simulating the next 20 days.

Precomputed E_start values:

Precomputed a_k = Estimate a(t₀-1)

Model plots

About the model

Recursive variant of SIR model where we drop the assumption that spreading decrease exponentially during a person's infectious period. We observe 5 compartments which people pass through: (S)usceptibles, (E)xposed but not infectious, (P)ropagating the virus, (I)nfected with symptoms, (R)emoved (i.e., recovery or death). We assume that every person stays a fixed period of time in the compartments E, P, and I. One might call it a delay model.

See our paper »Proposal of a recursive compartment model of epidemics and applications to the Covid-19 pandemic« at arxiv.org.

Input:

• Empirical data (C)onfirmed, (R)ecovered, (D)eaths by JHU humandata for several countries.
• List of intervals A_i (defined by their start points t_i) where parameter a(i) is constant.
• Constant parameters N, e, p, and q.
• Values for a(k) and E_add(k) computed from empirical data.

Output:

• Values for a(k)
• Model curves C = I_tot, I, and I_new.
• Root mean square error (RMSE) of Model curves.