Example model parameters and hyperparameters for priors — example

The example_params.toml file contains modifiable model inputs that get passed to stan to fit the model. The parameters used here have been chosen by the authors to best reflect SARS-CoV-2 infection, hospital admissions, and wastewater components. See the Model Parameters section of the model_definition.md in our Github for more information/justification of these prior/parameter choices.

If running the model on SARS-CoV-2 hospital admissions and wastewater concentrations, it may be reasonable to use these example parameters as a starting place for your data (aka none are dataset specific). The values that will need tweaking for a specific dataset are defined in the documentation for example_df.

example_params <- cfaforecastrenewalww::get_params(
  fs::path_package(
    "extdata", "example_params.toml",
    package = "cfaforecastrenewalww"
  )
)

Format

A .toml file that gets formatted as a 1 row dataframe where each column is a parameter name and its value is indicated in the first row

uot: "unoubserved time" an integer that dictates the duration of time (in days) that the model is initialized with exponential growth before any observations are available.
r_prior_mean: Prior on mean R(t)
r_prior_sd: Prior on standard deviation in the mean R(t)
sigma_rt_prior: Prior on the standard deviation between subpopulation level R(t)s
i0_certainty: Prior on the certainty in initial infections I0/N, representing the effective number of binomial trials in the beta prior centered on I0/N
initial_growth_prior_mean: Prior on mean exponential growth rate(daily) in the unobserved time during infection initialization
initial_growth_prior_sd: Prior on the standard deviation in the daily exponential growth rate during the unobserved time
autoreg_rt_a: Prior on the autoregulation term on the R(t) trend and the AR(1) process in subpopulation level R(t), shape1 parameter of a beta prior
autoreg_rt_b: Prior on the autoregulation term on the R(t) trend and the AR(1) process in subpopulation level R(t), shape2 parameter of a beta prior
eta_sd_sd: Prior on standard deviation in the deviation in the R(t) trend
infection_feedback_prior_logmean: Prior on the log mean infection feedback term
infection_feedback_prior_logsd: Prior on log standard deviation of the infection feedback term)
p_hosp_mean: Prior on the mean infection hospital admissions rate
p_hosp_sd_logit: Prior on the logit scale of the standard deviation of the mean infection hospital admissions rate
p_hosp_sd_sd: Prior on the standard deviation in the time-varying deviation in the infection hospital admissions rate
inv_sqrt_phi_prior_mean: Prior on the mean of the inverse square root of the negative binomial phi of the hospital admissions observation process
inv_sqrt_phi_prior_sd: Prior on the standard deviation of the inverse square root of the negative binomial phi of the hospital admissions observation process
wday_effect_prior_mean: Prior on the mean amount of weight assigned to each day of the week proportional to the whole week
wday_effect_prior_sd: Prior on the standard deviation in the amount of weight assigned to each day of the week proportional to the whole week
ml_of_ww_per_person_day: Set value of the estimated number of mL of wastewter produced per person per day
t_peak_mean: Prior on the mean time of peak fecal shedding in wastewater
t_peak_sd: Prior on the standard deviation in the time of peak fecal shedding in wastewater
viral_peak_mean: Prior on mean peak viral load (in log10 scale) of fecal shedding of viral genomes in wastewaster
viral_peak_sd: Prior on the standard deviation in the peak viral load (in log10 scale) of fecal shedding of viral genomes in wasteawter
duration_shedding_mean: Prior on mean duration of fecal shedding in wastewater
duration_shedding_sd: Prior on the standard deviation of fecal shedding in wastewater
log10_g_prior_mean: Prior on the mean number of genomes shed per infected individual in log10 scale
log10_g_prior_sd: Prior on standard deviation in the number of genomes shed per infected individual in log10 scale
log_g_prior_mean: Prior on the mean number of genomes shed per infected individual in log scale
log_g_prior_sd: Prior on standard deviation in the number of genomes shed per infected individual in log scale
sigma_ww_site_prior_mean_mean: Prior on mean of the mean site-lab-level observation error
sigma_ww_site_prior_mean_sd: Prior on standard deviation of the mean site-lab-level observation error
sigma_ww_site_prior_sd_mean: Prior on the mean of the standard deviation across sites in site-lab-level observation errors
sigma_ww_site_prior_sd_sd: Prior on standard deviaiton of the standard deviation across sites in site-lab level observation errors
ww_site_mod_sd_sd: Prior on the standard deviation in the standard deviation between site-lab level multipliers
log_phi_g_prior_mean: Prior on mean individual level dispersion in number of genomes shed per infection, not currently used in the model
log_phi_g_prior_sd: Prior on the standard deviation in individual level dispersion in number of genomes shed per infection, not currently used in the model
mu_gi: Set parameter for the log mean of the generation interval
sigma_gi: Set parameter for the log standard deviaiton of the generation interval
gt_max: Integer indicating the maximum number of days for which secondary transmission can occur
r: Exponential growth rate on the correction factor applied to the incubation period estimate
backward_shape: Set parameter for the backward shape parameter of the Weibull distribution for the incubation period
backward_scale: Set parameter for the backward scale parameter of the Weibull distribution for the incubation period
neg_binom_mu: Set parameter for the mean of the negative binomially distributed time from symptom onset to hospital admission
neg_binom_size: Set parameter for the size parameter of the negative binomially distributed time from symptom onset to hospital admissions

Source

The set parameters for the generation interval and incubation period were obtained from: Park, Sang Woo, et al. "Inferring the differences in incubation-period and generation-interval distributions of the Delta and Omicron variants of SARS-CoV-2." Proceedings of the National Academy of Sciences 120.22 (2023): e2221887120.

See Model Parameters for the full set of references for each prior.

Details

These parameters specify the priors on key parameters, except where indicated as set parameters. In these instances (generation interval, incubation period, symptom onset to hospital admissions) we pre-define the delay distrubions using estimates from the literature.