Read on Twitter
This @Accad_Koka podcast with @mgmgomes1 on heterogeneity / herd immunity is a MUST-listen. THREAD on interesting takeaways. To start with, Dr. Gomes is an expert and hardly a "COVID denier" - she actually supports spring lockdowns - yet still firmly believes HIT ~20%. (1/n) https://twitter.com/Accad_Koka/status/1295013102862258177
Dr. Gomes has 30 years of research experience in nonlinear dynamics, and has focused on epidemiology for 20 years, specifically heterogeneity over the past 10 years. Here is her brief CV. (2/n)
She has been working on heterogeneity in other diseases such as tuberculosis and malaria for the past decade. Ex https://royalsocietypublishing.org/doi/full/10.1098/rspb.2011.2712 : "By accounting for heterogeneity , we obtain significantly better model fittings to epidemiological data of TB transmission." (3/n)
Early on (April), there was not enough data for her team to determine the "coefficient of variation" (degree of heterogeneity) for COVID-19. So they just provided various scenarios based on other similar diseases. (4/n)
Reopening, however, provided them with substantially more data. Homogeneous (Osterholm/Ferguson/Hotez) models would predict exponential growth after mitigation measures were relaxed. That's not what actually happened in hard-hit areas in Europe and the U.S. (5/n)
Based on this data, they were able to estimate the coefficient of variation, understanding the difference in heterogeneity (as well as R0) in different areas. Gomes observes NYC had much higher R than most areas. (6/n)
Despite the fact that conventional 1 - (1/R0) homogeneous SIR models were not fitting the data, and the Gomes team's heterogeneity models were, most epidemiologists continued to insist on using the classical models. (7/n)
Essentially, epidemiologists refuse to engage constructively with Gomes - they offer vague, non-data-driven criticisms such as "I doubt that's true" or "we don't agree" - without explaining where or why her math is wrong, or how else to explain real world cases like Sweden. (8/n)
If you listen to the podcast, you'll clearly hear that Gomes is just a researcher - she has no political agenda and, in fact, supports many of the interventions that have been taken - she's just trying to look at the math and do the data. And she's been ignored. (9/n)
Heterogeneity self-obviously exists on two levels. First, connectivity - some of us are more connected than others. Second, susceptibility - COVID data makes it completely obvious that not all of us are equally susceptible. (10/n)
Gomes is frustrated her work isn't being taken seriously despite representing the best current model we have of pandemic behavior - the other data scientists with the best track records here like @youyanggu agree that there is an immunity component at play in the data. (11/n)
Anyway, the podcast is well worth listening to. Al of us need to get the word out about how Dr. Gomes's model is more accurately predicting real-time epidemic dynamics in areas like the Sun Belt than the failed Osterholm/Hotez/Ferguson homogeneous models. (12/n)
Dr. Gomes welcomes actual data-driven pushback - she's just not getting any from people who are actually doing the math, rather than just spouting opinions. Please follow her on Twitter ( @mgmgomes1) and retweet her posts - her contributions are valuable. (13/n)
Real-world takeaway summary is what @youyanggu has been observing here: at least so far, we're only seeing big "second peaks" in areas that didn't have a meaningful first peak to begin with - even within same state/country, such as LA/Idaho/Spain. (n/n) https://twitter.com/youyanggu/status/1294324998677639170
