github地址

# SIR model on Network

We here reconsider the SIR model in network where k k is the number of neighbor and v ( t ) v(t) is the average number of infected neighbor, v v is the average ratio of infected neighbor, w ( t ) w(t) is the average number of neighbors recovered. We have:

d s k d t = − β k v s k d i k d t = β k v s k − γ i k d r k d t = γ i k v ( t ) = ∑ k = 0 ∞ q k i k ( t ) w ( t ) = ∑ k = 0 ∞ q k r k ( t ) T a k e   t h e   d e r i v a t i v e   o f   w ( t ) = ∑ k = 0 ∞ q k r k ( t ) d w d t = ∑ k = 0 ∞ q k d r k d t = γ ∑ k = 0 ∞ q k i k = γ v ( t ) ∴ v ( t ) = 1 γ d w d t d s k d t = − β k v s k ⇔ d s k d t = − β k 1 γ d w d t s k \frac{ds_k}{dt}=-\beta kvs_k\\ \frac{di_k}{dt}=\beta kvs_k-\gamma i_k\\ \frac{dr_k}{dt}=\gamma i_k\\ v(t)=\sum_{k=0}^\infin q_k i_k(t)\\ w(t)=\sum_{k=0}^\infin q_k r_k(t)\\ Take~the~derivative~of~ w(t)=\sum_{k=0}^\infin q_k r_k(t)\\ \frac{dw}{dt}=\sum_{k=0}^\infin q_k \frac{dr_k}{dt}=\gamma \sum_{k=0}^\infin q_ki_k=\gamma v(t)\\ \therefore v(t)=\frac{1}{\gamma}\frac{dw}{dt}\\ \frac{ds_k}{dt}=-\beta kvs_k\Leftrightarrow \frac{ds_k}{dt}=-\beta k\frac{1} {\gamma}\frac{dw}{dt}s_k\\
We could solve these equation and get:

s k = s 0 e − β γ k w = s 0 [ e − β γ w ] k = s 0 [ u ( t ) ] k   w h e r e   u ( t ) = e − β γ w   w ( t ) = − γ β l n u s_k=s_0e^{-\frac{\beta}{\gamma}kw}\\ =s_0[e^{-\frac{\beta}{\gamma}w}]^k\\ =s_0[u(t)]^k~\\ where~u(t)=e^{-\frac{\beta}{\gamma}w}\\ ~w(t)=-\frac{\gamma}{\beta}ln u\\
With s + i + r = 1 s+i+r=1 , we could drive:

v ( t ) = ∑ k = 0 ∞ q k i k ( t ) = ∑ k = 0 ∞ q k ( 1 − r k − s k ) = 1 − w ( t ) − s 0 ∑ k = 0 ∞ q k u k = 1 + γ β l n u − s 0 G 1 ( u ) ∵ w ( t ) = − γ β l n u d w d t = γ v ( t ) ⇔ − γ β 1 u d u d t = − γ [ 1 + γ β l n u − s 0 G 1 ( u ) ] ∴ d u d t = − β u [ 1 + γ β l n u − s 0 G 1 ( u ) ] v(t)=\sum_{k=0}^\infin q_k i_k(t)\\ =\sum_{k=0}^\infin q_k(1-r_k-s_k)\\ =1-w(t)-s_0\sum_{k=0}^\infin q_ku^k\\ =1+\frac{\gamma}{\beta}ln u-s_0G_1(u)\\ \because w(t)=-\frac{\gamma}{\beta}ln u\\ \frac{dw}{dt}=\gamma v(t)\Leftrightarrow -\frac{\gamma}{\beta}\frac{1}{u}\frac{du}{dt}=-\gamma[1+\frac{\gamma}{\beta}ln u-s_0G_1(u)]\\ \therefore \frac{du}{dt}=-\beta u[1+\frac{\gamma}{\beta}ln u-s_0G_1(u)]

Set u = 1 − ϵ u=1-\epsilon we could get

d ϵ d t = [ β G 1 ′ ( 1 ) − γ ] ϵ a s s u m e   s 0 = 1 u ( t ) = 1 − ϵ ( t ) = 1 − e ( β G 1 ′ ( 1 ) − γ ) t s k ( t ) = u k = [ 1 − e ( β G 1 ′ ( 1 ) − γ ) t ] k ≃ 1 − e k ( β G 1 ′ ( 1 ) − γ ) t \frac{d\epsilon }{dt}=[\beta G_1'(1)-\gamma]\epsilon \\ assume~s_0=1\\ u(t)=1-\epsilon (t)=1-e^{(\beta G_1'(1)-\gamma)t}\\ s_k(t)=u^k=[1-e^{(\beta G_1'(1)-\gamma)t}]^k\simeq 1-e^{k(\beta G_1'(1)-\gamma)t}\\

S e t   s 0 = 1 d u d t = − β u [ 1 + γ β l n u − G 1 ( u ) ] s + i + r = 1 s = u k Set~s_0=1\\ \frac{du}{dt}=-\beta u[1+\frac{\gamma}{\beta}ln u-G_1(u)]\\ s+i+r=1\\ s=u^k

With the above equations, we could fit the lines of the rate of susceptible, infectious and recovered people and get the parameter estimation of γ \gamma and β \beta . ( γ \gamma in the MGF method is just parameter for τ i \tau_i , since p ( τ ) d τ = γ e γ τ d τ p(\tau)d\tau=\gamma e^{\gamma \tau}d\tau , β \beta is the T i j T_{ij} which is simplified for the paired transmission)

If we set s k ( t ) = 0 s_k(t)=0 , we could get
β γ = 1 G 1 ′ ( 1 ) \frac{\beta}{\gamma}=\frac{1}{G_1'(1)}\\
Thus, average outbreak size here is β γ \frac{\beta}{\gamma} .

Here, we could rewrite the model of SIR on network:
s k ( t ) = u ( t ) k d u d t = − β u [ 1 + γ β l n u − s 0 G 1 ( u ) ] u ( t ) = 1 − e ( β G 1 ′ ( 1 ) − γ ) t d i k d t = k u ( t ) k − 1 d u d t − γ i k d r k d t = γ i k s + i + r = 1 s_k(t)=u(t)^k\\ \frac{du}{dt}=-\beta u[1+\frac{\gamma}{\beta}ln u-s_0G_1(u)]\\ u(t)=1-e^{(\beta G_1'(1)-\gamma)t}\\ \frac{di_k}{dt}=ku(t)^{k-1}\frac{du}{dt}-\gamma i_k\\ \frac{dr_k}{dt}=\gamma i_k\\ s+i+r=1\\
Also, we could drive their iteration form with S = s N , I = i N , R = r N S=sN, I=iN, R=rN and let c c be the number of people one contact:
U n = U n − 1 − β U n − 1 [ 1 + γ β l n U n − 1 − s 0 G 1 ( U n − 1 ) ] S n = U n k I n = I n − 1 − k U n − 1 ( t ) k β U n − 1 [ 1 + γ β l n U n − 1 − s 0 G 1 ( U n − 1 ) ] − γ I n − 1 R n = R n − 1 + γ I n − 1 U_n=U_{n-1}-\beta U_{n-1}[1+\frac{\gamma}{\beta}ln U_{n-1}-s_0G_1(U_{n-1})]\\ S_n=U_n^k\\ I_n=I_{n-1}-kU_{n-1}(t)^k\beta U_{n-1}[1+\frac{\gamma}{\beta}ln U_{n-1}-s_0G_1(U_{n-1})]-\gamma I_{n-1}\\ R_n=R_{n-1}+\gamma I_{n-1}\\

# SEIR model on Network

We here reconsider the SEIR model in network where k k is the number of neighbor and v ( t ) v(t) is the average number of infected neighbor, v v is the average ratio of infected neighbor, m ( t ) m(t) is the average number of exposed neighbor, w ( t ) w(t) is the average number of neighbors recovered. We have:

d s k d t = − β k v s k d e k d t = β k v s k − α e k d i k d t = α e k − γ i k d r k d t = γ i k v ( t ) = ∑ k = 0 ∞ q k i k ( t ) m ( t ) = ∑ k = 0 ∞ q k e k ( t ) w ( t ) = ∑ k = 0 ∞ q k r k ( t ) T a k e   t h e   d e r i v a t i v e   o f   w ( t ) = ∑ k = 0 ∞ q k r k ( t ) d w d t = ∑ k = 0 ∞ q k d r k d t = γ ∑ k = 0 ∞ q k i k = γ v ( t ) ∴ v ( t ) = 1 γ d w d t d s k d t = − β k v s k ⇔ d s k d t = − β k 1 γ d w d t s k T a k e   t h e   d e r i v a t i v e   o f   m ( t ) = ∑ k = 0 ∞ q k e k ( t ) d m ( t ) d t = ∑ k = 0 ∞ q k d e k d t = β k v ∑ k = 0 ∞ q k s k − ∑ k = 0 ∞ q k α e k = β k v ∑ k = 0 ∞ q k s k − α m ( t ) T a k e   t h e   d e r i v a t i v e   o f   v ( t ) = ∑ k = 0 ∞ q k i k ( t ) d v ( t ) d t = ∑ k = 0 ∞ q k d i k d t = α m ( t ) − γ v ( t ) m ( t ) = 1 α [ d v ( t ) d t + γ v ( t ) ] \frac{ds_k}{dt}=-\beta kvs_k\\ \frac{de_k}{dt}=\beta kvs_k-\alpha e_k\\ \frac{di_k}{dt}=\alpha e_k-\gamma i_k\\ \frac{dr_k}{dt}=\gamma i_k\\ v(t)=\sum_{k=0}^\infin q_k i_k(t)\\ m(t)=\sum_{k=0}^\infin q_k e_k(t)\\ w(t)=\sum_{k=0}^\infin q_k r_k(t)\\ Take~the~derivative~of~ w(t)=\sum_{k=0}^\infin q_k r_k(t)\\ \frac{dw}{dt}=\sum_{k=0}^\infin q_k \frac{dr_k}{dt}=\gamma \sum_{k=0}^\infin q_ki_k=\gamma v(t)\\ \therefore v(t)=\frac{1}{\gamma}\frac{dw}{dt}\\ \frac{ds_k}{dt}=-\beta kvs_k\Leftrightarrow \frac{ds_k}{dt}=-\beta k\frac{1}{\gamma}\frac{dw}{dt}s_k\\ Take~the~derivative~of~ m(t)=\sum_{k=0}^\infin q_k e_k(t)\\ \frac{dm(t)}{dt}=\sum_{k=0}^\infin q_k \frac{de_k}{dt}=\beta k v \sum_{k=0}^\infin q_ks_k-\sum_{k=0}^\infin q_k \alpha e_k\\ =\beta k v \sum_{k=0}^\infin q_ks_k-\alpha m(t)\\ Take~the~derivative~of~ v(t)=\sum_{k=0}^\infin q_k i_k(t)\\ \frac{dv(t)}{dt}=\sum_{k=0}^\infin q_k \frac{di_k}{dt}= \alpha m(t)-\gamma v(t)\\ m(t)=\frac{1}{\alpha}[\frac{dv(t)}{dt}+\gamma v(t)]\\

We could solve these equation and get:

s k = s 0 e − β γ k w = s 0 [ e − β γ w ] k = s 0 [ u ( t ) ] k   w h e r e   u ( t ) = e − β γ w   w ( t ) = − γ β l n u s_k=s_0e^{-\frac{\beta}{\gamma}kw}\\ =s_0[e^{-\frac{\beta}{\gamma}w}]^k\\ =s_0[u(t)]^k~\\ where~u(t)=e^{-\frac{\beta}{\gamma}w}\\ ~w(t)=-\frac{\gamma}{\beta}ln u\\

With s + e + i + r = 1 s+e +i+r=1 , we could drive:

v ( t ) = ∑ k = 0 ∞ q k i k ( t ) = ∑ k = 0 ∞ q k ( 1 − r k − e k − s k ) = 1 − w ( t ) − m ( t ) − s 0 ∑ k = 0 ∞ q k u k = 1 + γ β l n u − m ( t ) − s 0 G 1 ( u ) ∴ m ( t ) = 1 + γ β l n u − s 0 G 1 ( u ) − v ( t ) ∵ m ( t ) = 1 α [ d v ( t ) d t + γ v ( t ) ] ∴ ( γ α + 1 ) v ( t ) + 1 α d v ( t ) d t = 1 + γ β l n u − s 0 G 1 ( u ) F o r   O D E   f u n c t i o n   d v d t + p ( t ) v = q ( t ) T h e   s o l u t i o n   i s   v ( t ) = e − ∫ p ( x ) d x C + e − ∫ p ( x ) d x ∫ q ( x ) e ∫ p ( x ) d x d x H e r e   p ( t ) = γ + α q ( t ) = [ 1 + γ β l n u − s 0 G 1 ( u ) ] α ∵ w ( t ) = − γ β l n u d w d t = γ v ( t ) ⇔ − γ β 1 u d u d t = − γ v ( t ) ∴ d u d t = β u v ( t ) = β u e − ∫ ( γ + α ) d x C + e − ∫ ( γ + α ) d x ∫ [ [ 1 + γ β l n u − s 0 G 1 ( u ) ] α ] e ∫ ( γ + α ) d x d x = C β u e − ( γ + α ) t + e − ( γ + α ) t ∫ [ [ 1 + γ β l n u ( t ) − s 0 G 1 ( u ( t ) ) ] α ] e ( γ + α ) t d t v(t)=\sum_{k=0}^\infin q_k i_k(t)\\ =\sum_{k=0}^\infin q_k(1-r_k-e_k-s_k)\\ =1-w(t)-m(t)-s_0\sum_{k=0}^\infin q_ku^k\\ =1+\frac{\gamma}{\beta}ln u-m(t)-s_0G_1(u)\\ \therefore m(t)=1+\frac{\gamma}{\beta}ln u-s_0G_1(u)-v(t)\\ \because m(t)=\frac{1}{\alpha}[\frac{dv(t)}{dt}+\gamma v(t)]\\ \therefore (\frac{\gamma}{\alpha}+1)v(t)+\frac{1}{\alpha}\frac{dv(t)}{dt}=1+\frac{\gamma}{\beta}ln u-s_0G_1(u)\\ For~ODE~function ~\frac{dv}{dt}+p(t)v=q(t)\\ The~solution~is~ v(t)=e^{-\int p(x)dx}C+e^{-\int p(x)dx}\int q(x)e^{\int p(x)dx}dx\\ Here~p(t)=\gamma+\alpha\\ q(t)=[1+\frac{\gamma}{\beta}ln u-s_0G_1(u)]\alpha\\\\ \because w(t)=-\frac{\gamma}{\beta}ln u\\ \frac{dw}{dt}=\gamma v(t)\Leftrightarrow -\frac{\gamma}{\beta}\frac{1}{u}\frac{du}{dt}=-\gamma v(t)\\ \therefore \frac{du}{dt}=\beta uv(t)\\ =\beta ue^{-\int (\gamma+\alpha)dx}C+e^{-\int(\gamma+\alpha)dx}\int [[1+\frac{\gamma}{\beta}ln u-s_0G_1(u)]\alpha]e^{\int (\gamma+\alpha)dx}dx\\ =C\beta ue^{-(\gamma+\alpha)t}+e^{-(\gamma+\alpha)t}\int [[1+\frac{\gamma}{\beta}ln u(t)-s_0G_1(u(t))]\alpha]e^{ (\gamma+\alpha)t}dt\\

Assume   s 0 = 1 ~s_0=1 :

d u d t = C β u e − ( γ + α ) t + e − ( γ + α ) t ∫ [ [ 1 + γ β l n u ( t ) − s 0 G 1 ( u ( t ) ) ] α ] e ( γ + α ) t d t s k ( t ) = u k s + e + i + r = 1 \frac{du}{dt}=C\beta ue^{-(\gamma+\alpha)t}+e^{-(\gamma+\alpha)t}\int [[1+\frac{\gamma}{\beta}ln u(t)-s_0G_1(u(t))]\alpha]e^{ (\gamma+\alpha)t}dt\\ s_k(t)=u^k\\ s+e +i+r=1

With the above equations, we could fit the lines of the rate of susceptible, infectious and recovered people and get the parameter estimation of γ \gamma , β \beta , and α \alpha . ( γ \gamma in the MGF method is just parameter for τ i \tau_i , since p ( τ ) d τ = γ e γ τ d τ p(\tau)d\tau=\gamma e^{\gamma \tau}d\tau , β \beta is the T i j T_{ij} which is simplified for the paired transmission, α \alpha is the rate that exposed people become sick)

Here, we could rewrite the model of SEIR on network:
s k ( t ) = u ( t ) k d u d t = C β u e − ( γ + α ) t + e − ( γ + α ) t ∫ [ [ 1 + γ β l n u ( t ) − s 0 G 1 ( u ( t ) ) ] α ] e ( γ + α ) t d t d e k d t = k u ( t ) k − 1 d u d t − α e k d i k d t = α e k − γ i k d r k d t = γ i k s + e + i + r = 1 s_k(t)=u(t)^k\\ \frac{du}{dt}=C\beta ue^{-(\gamma+\alpha)t}+e^{-(\gamma+\alpha)t}\int [[1+\frac{\gamma}{\beta}ln u(t)-s_0G_1(u(t))]\alpha]e^{ (\gamma+\alpha)t}dt\\ \frac{de_k}{dt}=ku(t)^{k-1}\frac{du}{dt}-\alpha e_k\\ \frac{di_k}{dt}=\alpha e_k-\gamma i_k\\ \frac{dr_k}{dt}=\gamma i_k\\ s+e+i+r=1\\
Also, we could drive their iteration form with S = s N , E = e N , I = i N , R = r N S=sN, E=eN,I=iN, R=rN and let c c be the number of people one contact:
U n − α γ β l n U n + α G 1 ( U n ) = U n − 1 − α γ β l n U n − 1 + α G 1 ( U n − 1 ) + C β U n − 1 e − ( γ + α ) S n = U n k E n = E n − 1 + k U n − 1 ( t ) k − 1 [ C β U n − 1 e − ( γ + α ) + [ γ β l n U n − 1 − γ β l n U n − 2 − s 0 α [ G 1 ( U n − 1 ) − G 1 ( U n − 2 ) ] − α E n − 1 I n = I n − 1 + α E n − 1 − γ I n − 1 R n = R n − 1 + γ I n − 1 U_n-\frac{\alpha\gamma}{\beta}ln U_{n}+\alpha G_1(U_n)=U_{n-1}-\frac{\alpha\gamma}{\beta}ln U_{n-1}+\alpha G_1(U_{n-1})+C\beta U_{n-1}e^{-(\gamma+\alpha)}\\ S_n=U_n^k\\ E_n=E_{n-1}+kU_{n-1}(t)^{k-1}[C\beta U_{n-1}e^{-(\gamma+\alpha)}+ [\frac{\gamma}{\beta}ln U_{n-1}-\frac{\gamma}{\beta}ln U_{n-2}-s_0\alpha[G_1(U_{n-1})-G_1(U_{n-2})]-\alpha E_{n-1}\\ I_n=I_{n-1}+\alpha E_{n-1}-\gamma I_{n-1}\\ R_n=R_{n-1}+\gamma I_{n-1}

.

• 点赞
• 评论
• 分享
x

海报分享

扫一扫，分享海报

• 收藏 4
• 打赏

打赏

天天学习的零柒贰幺

你的鼓励将是我创作的最大动力

C币 余额
2C币 4C币 6C币 10C币 20C币 50C币
• 举报
• 一键三连

点赞Mark关注该博主, 随时了解TA的最新博文