Some Interesting Modeling Studies

I have generally not bothered posting on modeling studies, but here are three that actually advance the approach to understanding the dynamics and course of this epidemic.

In this first paper the authors took issue with the common approach of assuming homogeneity in a population.  (Science Article)   They created a model based on age structure and activity levels and showed how incorporating those can lower the level at which a pathogen has significantly reduced transmission.  They created six age groups with differing contact interactions and three activity levels, to which they arbitrarily assigned the population.  They also incorporated mitigation of spread measures.  Incorporating the activity levels had more of an impact on the level of population immunity than did incorporating age structure, but both had an effect.  Population immunity is generally projected at 60% to 70% of the population for CV but this version of the model has it occurring as low as 43%.  The authors recognize that there are many other factors that could and should be incorporated into a model to accurately reflect heterogeneity of the population.  And they note that their results indicate that more moderate mitigation of spread tactics resulted in fewer long-run cases and deaths than did severe ones.

This paper also focussed on the importance of including population variation.   (Medrxiv Paper)    They look more at variation in contagiousness, which they appear to attribute to variability in hygiene and social distancing, rather than innate characteristics.  They note in particular that there appear to be superspreaders who account for many transmissions of the virus.  They created three scenarios–a homogenous population, one in which 20% of the population is responsible for transmission in 80% of the cases and one in which 10% of the population is responsible for 80% of the spread.  With no intervention, in their homogenous model, 95% of the population becomes infected.   In the 20% heterogenous case, only 27.6% of the people become infected and only 14.1% in the 10% heterogenous case.  The heterogenous cases also have an earlier peak.  Their model also shows that early severe lockdowns result in worse epidemics than more measured approaches to mitigation of spread.

The third study similar assumed that this epidemic must not follow a typical pattern.   (Medrxiv Paper)   They also address the need to represent a heterogenous population in a model.  Unfortunately, their work involves extremely complex equations and is almost impossible for a non-PhD mathematician to follow.  But they claim that their revised model with heterogenous population interaction more closely approximates the actual experience of many countries.

While I appreciate the recognition of some modelers that variability in the population needs to be addressed, these particular efforts appear to miss the role of innate susceptibility, that is, there is a percent of the population which appears to have pre-existing immune defenses, and none of them adequately addresses the clustering problem worldwide in which the frail elderly, particularly in congregate care settings, are much more susceptible and represent the overwhelming number of severe illnesses and deaths.  Now that we know that, shouldn’t be too hard to build a model that replicates those features.