1

This system of differential equations corresponds to a compartmental model in epidemiology called SIRD:

\begin{aligned}&{\frac {dS}{dt}}=-{\frac {\beta IS}{N}}\\[6pt]&{\frac {dI}{dt}}={\frac {\beta IS}{N}}-\gamma I-\mu I\\[6pt]&{\frac {dR}{dt}}=\gamma I\\[6pt]&{\frac {dD}{dt}}=\mu I\end{aligned}

Note that:

\begin{aligned}&{\frac {dS}{dt}}+{\frac {dI}{dt}}+{\frac {dR}{dt}}+{\frac {dD}{dt}}=0\end{aligned}

Therefore:

\begin{aligned}&{\displaystyle S(t)+I(t)+R(t)+D(t)=N}\end{aligned}

If we set μ equal to zero, it simplifies to the SIR model:

\begin{aligned}&{\frac {dS}{dt}}=-\beta IS\\[6pt]&{\frac {dI}{dt}}=\beta IS-\gamma I\\[6pt]&{\frac {dR}{dt}}=\gamma I\end{aligned}

Note that:

\begin{aligned}&{\frac {dS}{dt}}+{\frac {dI}{dt}}+{\frac {dR}{dt}}=0\end{aligned}

Therefore:

\begin{aligned}&{\displaystyle S(t)+I(t)+R(t)=N}\end{aligned}

It can be further simplified to the SI model:

\begin{aligned}{\frac {dS}{dt}}&=-{\frac {\beta SI}{N}}+\gamma I\\[6pt]{\frac {dI}{dt}}&={\frac {\beta SI}{N}}-\gamma I\end{aligned}

Note that:

\begin{aligned}&{\frac {dS}{dt}}+{\frac {dI}{dt}}=0\end{aligned}

Therefore:

\begin{aligned}&{\displaystyle S(t)+I(t)=N}\end{aligned}

Its exact solution is a logistic function.

This is a logistic function:

\begin{aligned}&{\displaystyle f(x)={\frac {L}{1+e^{-k(x-x_{0})}}}}\end{aligned}

This is the exact solution:

\begin{aligned}&{\displaystyle I(t)={\frac {I_{\infty }}{1+Ve^{-\chi t}}}}\end{aligned}

\begin{aligned}&{\displaystyle I_{\infty }=(1-\gamma /\beta )N}\end{aligned}

\begin{aligned}&{\displaystyle \chi =\beta -\gamma }\end{aligned}

\begin{aligned}&{\displaystyle V=I_{\infty }/I_{0}-1}\end{aligned}

\begin{aligned}&{\displaystyle S(t)=N-I(t)}\end{aligned}

I conjectured that approximate solutions for the SIRD and SIR could involve a Gompertz function.

This is a formula for the Gompertz function:

\begin{aligned}&{\displaystyle f(t)=a\mathrm {e} ^{-b\mathrm {e} ^{-ct}}}\end{aligned}

And this is an alternative formula:

\begin{aligned}&{\displaystyle N(t)=N_{0}\exp(\ln(N_{\infty }/N_{0})(1-\exp(-bt)))}\end{aligned}

t is time

N_0 is the initial density of cells

N_inf is the plateau cell/population density

b is the initial rate of tumor growth

Specifically, I conjectured that:

For the SIRD, the approximations are:

  • S as N minus a Gompertz function
  • R as a Gompertz function
  • D as a Gompertz function

For the SIR model:

  • S as N minus a Gompertz function
  • R as a Gompertz function

I would like to prove mathematically that the approximations are correct within some margin of error. Do I just add an O(t) to the function and plug in the system of equations and see what margin of error comes out?

  • Sorry there was a typo. If we derive it correctly then we get $(\log S)'' = (\beta S - \gamma)(\log S)'$. So it was not as exact as that, but atleast we can see that a double exponential like behavior is expected in some regimes. – Winther Oct 12 '24 at 09:56
  • @Winther thank you. this comment is very helpful. could you elaborate it into an answer please? I am still at a loss on how to prove the magnitude of the error of my approximation is small, though. –  Oct 12 '24 at 13:09
  • Have you isolated the differential equations that are satisfied by a Gompertz function. Can you formulate a suitable system differential equations that is satisfied by the difference between the SIR solution and a parameterized family of Gompertz functions? The goal is to obtain an initial value problem with zero as the initial condition and a right-hand side that you can control. – Carl Christian Oct 13 '24 at 11:06

0 Answers0