Later in today's blog I will refer to the modeling done at Univ. Washington using Gaussian curves of the Daily Death Count. The UW model is used by governors for local hospital utilization planning. Dr. Birx referred to it in the March 31 White House briefing.
The latest Westchester data from March 31.
New Cases compared with NYC and Nassau County with new data from March 31.
The full report is here: http://www.healthdata.org/sites/default/files/files/research_articles/2020/COVID-forecasting-03252020_4.pdf
Daily Updated Projections by state are here:
https://covid19.healthdata.org/projections
The Institute for Health Metrics and Evaluation (IHME), an independent global health research center at the University of Washington, has constructed a state-by-state model for COVID-19 disease spread. It appears to be the model of choice for governors preparing for the next few weeks. It has been mentioned by name in a few of the White House Task Force Briefings.
The model predicts a daily death count for each state. That number is used to make other predictions, including daily hospital utilization.
The daily death count for each state is a 'bell (gaussian) curve' where the height and width are determined by the date and extent of implementation of 4 mitigation strategies. The curve for New York is shown above.
Below are the authors' disclaimers. My comments are in brackets.
"Any attempt to forecast the COVID-19 epidemic has many limitations. Only one location has had a generalized epidemic and has currently brought new cases to 0 or near 0 [in other words, only Wuhan has a fully shaped curve for model validation], namely Wuhan. Many other locations, including all other provinces in China, have so far successfully contained transmission, preventing a general outbreak [hence those provinces were not used to test the model because they did not follow a Bell Curve?] Modeling for US states based on one completed epidemic, at least for the first wave, and many incomplete epidemics is intrinsically challenging.
The consequent main limitation of our study is that observed epidemic curves for COVID-19 deaths [?there exists only one complete epidemic curve for COVID-19] define the likely trajectory for US states. In this study, we do include a covariate meant to capture the timing of social distancing measures to take into account that Wuhan implemented 4 out of 4 social distancing measures within 6 days of reaching a threshold death rate of 0.31 per million. Our models explicitly take into account variation in age-structure, which is a key driver of all-age mortality. But these efforts at quantification do not take into account many other factors that may influence the epidemic trajectory: the prevalence of chronic lung disease, the prevalence of multi-morbidity, population density, use of public transport, and other factors that may influence the immune response. We also have not explicitly incorporated the effect of reduced quality of care due to stressed and overloaded health systems beyond what is captured in the data. For example, the higher mortality rate in Italy is likely in part due to policies around restricting invasive ventilation in the elderly. The model ensemble used does suggest that locations with faster increases in the death rate are likely to have more peak case load and cumulative deaths, but our uncertainty intervals are appropriately large."
One source of error sticks out to me; the model assumes that mitigation measure #4 is not important. Wuhan implemented this measure, but no US state has implemented it. Measure #4 is severe limitations on travel, such as shutting down public transportation and restricting travel in and out of Wuhan to essential needs. From the paper: "Days with 1 measure were counted as 0.67 equivalents, days with 2 measures as 0.334 equivalents and with 3 or 4 measures as 0."
No comments:
Post a Comment