A new design of an adaptive model of infectious diseases based on artificial intelligence approach: monitoring and forecasting of COVID-19 epidemic cases

Nail, Bachir; Rabehi, Abdelaziz; Bekhiti, Belkacem; Arbaoui, Taha

Background: Mathematical infectious disease models available in literature, mostly take in their design that the parameters of basic reproduction number R_0 and interval serial S_I as constant values during tracking the outbreak cases. In this report a new intelligent model called HH-COVID-19 is proposed, with simple design and adaptive parameters. Methods: The parameters R_0 and S_I are adapted by adding three new weighting factors α, β and γ and two free parameters σ_1 and σ_2 in function of time t, thus the HH-COVID-19 become time-variant model. The parameters R_0, S_I, α, β, γ, σ_1 and σ_2 are estimated optimally based on a recent algorithm of artificial intelligence (AI), inspired from nature called Harris Hawks Optimizer (HHO), using the data of the confirmed infected cases in Algeria country in the first t=55 days. Results: Parameters estimated optimally: R_0= 1.341, S_I= 5.991, α= 2.987, β= 1.566, γ= 4.998 σ_1= -0.133 and σ_2= 0.0324. R_0 starts on 1.341 and ends to 2.677, and S_I starts on 5.991 and ends to 6.692. The estimated results are identically to the actual infected incidence in Algeria, HH-COVID-19 proved its superiority in comparison study. HH-COVID-19 predicts that in 1 May, the infected cases exceed 50 000, during May, to reach quickly the herd immunity stage at beginning of July. Conclusion: HH-COVID-19 can be used for tracking any COVID-19 outbreak cases around the world, just should updating its new parameters to fitting the area to be studied, especially when the population is directly vulnerable to COVID-19 infection.