George Rebane
As promised, I finally finished the V1.0 version of Epidyne during this enforced C19 self-quarantine, or hunkering down period as it is better known. The stay at home has allowed a lot of things to get accomplished on everyone’s ‘do list’. At our house we have not had a boring moment – so many interesting projects, and now some uninterrupted time to do them. For me, developing a working epidemic spread model has been a lot of fun and also very consuming. I hope the interested readers will also be stimulated by and gain a better understanding of the terrible C19 pandemic now sweeping the world wreaking havoc on lives and livelihoods.
Motivation. In light of the daily reported dilemmas of widely divergent epidemic spread models that are being consulted by our national C19 response planners, and also the background reports of their various vintages and provenances that make them all more or less difficult to use, let alone upgrade or even maintain, I decided to develop my own version so that I could play out various recorded data and policy alternatives, and inform RR readers of results I considered insightful, informative, or at least interesting.
Objective. My objective was to develop a model that was sufficiently complex to simulate and predict realworld experiences and outcomes, but no more than that. The model had to have an adequately rich input space (think of them as control knobs) that could be set to represent both policy alternatives (such as time/size dependent quarantines, and regional populations), and also incorporate field measured data (such as infection rates dependent on virulence and vectors, contagion durations, and onset time lags). Most certainly the model had to be able to handle the critical infection rate dynamics driven by the process known as ‘herd immunity’.
RR readers got an early peek at Epidyne, my home-brew epidemiological spread model, in the last ‘Hunker Down Diary’ post (here). Since then I’ve finished testing it and fixed the inevitable collection of bugs. In this post I will do a first cut intro to describe the model and demonstrate its workings through a few scenarios. Epidyne has been programmed in both an Excel™ spreadsheet and the system development language Matlab™. A portion of the spreadsheet model is shown below.
Epidyne models the time history of the sizes and change rates (time derivatives) of four inter-related population cohorts that comprise a regional population into which an infectious disease is introduced that has the potential of becoming an epidemic. These cohorts are comprised of pU - those uninfected yet vulnerable; pI - those currently infected; pR - those who have recovered and are no longer infectious; and pD - the deceased. pU may also be mediated by timed quarantines which can reduce the size of pU or increase it when the quarantines end or are (partially) ignored before the epidemic runs its course.
As shown above, the population flows from vulnerable uninfected cohort to the infected cohort, which then feeds the non-decreasing recovered and the deceased cohorts. The asymptomatic and pre-symptomatic infected continue to reduce the uninfected population at a variable ‘reproduction’ rate rI (cf. herd immunity). The course of the disease in an infected individual is completed within nI weeks after which the individual has either recovered or died. The mortality rate of the infected is given by rD.
The above is from the technical note that describes Epidyne and illustrates its use. It may be downloaded here. Download TN2004-1_Epidyne
Comments