Selected article for: "exponential growth rate and total number"

Author: Rodolfo Jaffe; Mabel Patricia Ortiz Vera; Klaus Jaffe
Title: Globalized low-income countries may experience higher COVID-19 mortality rates
  • Document date: 2020_4_3
  • ID: 1ntplgl6_11
    Snippet: To assess country-level COVID-19 infection rates we fitted generalized linear models with a negative binomial distribution of errors (and a logarithmic link function) to account for overdispersion, using the glm.nb function from the MASS R package (Venables & Ripley, 2002) . The daily number of confirmed cases was used as the response variable and date as the predictor. The model's coefficient for date was then used as a proxy for infection rate .....
    Document: To assess country-level COVID-19 infection rates we fitted generalized linear models with a negative binomial distribution of errors (and a logarithmic link function) to account for overdispersion, using the glm.nb function from the MASS R package (Venables & Ripley, 2002) . The daily number of confirmed cases was used as the response variable and date as the predictor. The model's coefficient for date was then used as a proxy for infection rate during the exponential growth-phase. Mortality rates for each country were calculated as (Total number of deaths / Total number of cases) x 100. We then excluded observations containing missing data (final sample size was 36 countries) and ran linear multiple regressions with either infection rate or mortality rate as response variables, and all possible combinations of up to three non-correlated predictor variables to avoid overfitting. All climate and socio-economic variables plus latitude were included as predictor variables. Models where constructed using the dredge function from the MuMIn package (Bartoń, 2019 ) and a custom script (https://github.com/rojaff/dredge_mc). We then selected the set of best-fitting models using the Akaike Information Criterion (ΔAIC AIC ≤ 2) and used them to calculate confidence Intervals (CI) for modelaveraged coefficients and sum of Akaike weights. All analyses were performed in R (see Script_S1).

    Search related documents:
    Co phrase search for related documents
    • Akaike Information Criterion and best fit model: 1, 2
    • Akaike Information Criterion and CI confidence interval: 1, 2, 3, 4
    • Akaike Information Criterion and confidence interval: 1, 2, 3, 4, 5, 6, 7, 8
    • Akaike Information Criterion and country level: 1
    • Akaike Information Criterion and infection rate: 1
    • Akaike Information Criterion and linear model: 1, 2, 3
    • Akaike Information Criterion and mortality rate: 1
    • Akaike Information Criterion and multiple regression: 1, 2
    • Akaike Information Criterion and total number: 1