Author: Spencer Woody; Mauricio Garcia Tec; Maytal Dahan; Kelly Gaither; Spencer Fox; Lauren Ancel Meyers; James G Scott
Title: Projections for first-wave COVID-19 deaths across the US using social-distancing measures derived from mobile phones Document date: 2020_4_22
ID: 87lxnslh_15
Snippet: The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101 /2020 in the original formulation of the model) can inform U.S. state-level forecasts. For a variety of reasons, we find this assumption problematic. Our forecasts therefore rely solely on U.S. data, with state-level parameters shrunk toward a common mean in a hierarchical Bayesian model. Difference 3: valid uncertainty quantification. We addres.....
Document: The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101 /2020 in the original formulation of the model) can inform U.S. state-level forecasts. For a variety of reasons, we find this assumption problematic. Our forecasts therefore rely solely on U.S. data, with state-level parameters shrunk toward a common mean in a hierarchical Bayesian model. Difference 3: valid uncertainty quantification. We address a problem with the IHME model by relying on a fundamentally different statistical assumption about model errors. Briefly: the IHME model fits cumulative death rates using a least-squares-like procedure on the log scale and applying standard large-sample statistical theory to get confidence intervals. For this procedure to result in valid uncertainty quantification, one must assume that successive model errors are independent. But in the IHME fitting procedure, this assumption is violated: today's cumulative death rate is yesterday's plus an increment, so the two must be correlated. Our model repairs this problem by fitting daily (noncumulative) death rates using a mixed-effects negative-binomial generalized linear model.
Search related documents:
Co phrase search for related documents- bayesian model and death rate: 1, 2, 3, 4, 5, 6, 7
- bayesian model and hierarchical bayesian model: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73
- bayesian model and IHME model: 1
- bayesian model and large sample: 1
- bayesian model and linear model: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38
- bayesian model and log scale: 1
- bayesian model and model error: 1, 2, 3, 4, 5, 6, 7, 8, 9
- bayesian model and negative binomial: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11
- bayesian model and problem address: 1
- bayesian model and state level: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
- bayesian model and state level forecast: 1
- bayesian model and uncertainty quantification: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12
- common mean and confidence interval: 1, 2, 3, 4
- common mean and linear model: 1
Co phrase search for related documents, hyperlinks ordered by date