nof1.nma.data: Make an N of 1 object containing data, priors, and a jags...

In jiabei-yang/nof1ins: N-of-1 study general analysis tool

Description

Make an N of 1 object containing data, priors, and a jags model file for (network) meta-analyses

Usage

nof1.nma.data(
  Y,
  Treat,
  baseline.Treat,
  ID,
  response,
  model.linkfunc = NULL,
  model.intcpt = "fixed",
  model.slp = "random",
  ord.ncat = NULL,
  ord.model,
  ord.parallel = NULL,
  strata.cov = NULL,
  adjust.strata.cov = NULL,
  lvl2.cov = NULL,
  spline.trend = F,
  trend.type,
  y.time = NULL,
  knots = NULL,
  trend.df = NULL,
  step.trend = F,
  y.step = NULL,
  corr.y = F,
  alpha.prior = NULL,
  beta.prior = NULL,
  eta.prior = NULL,
  dc.prior = NULL,
  c1.prior = NULL,
  rho.prior = NULL,
  hy.prior = NULL,
  na.rm = T,
  ...
)

Arguments

Y	Outcome of the study. This should be a vector with NA's included in time order.
Treat	Treatment indicator vector with the same length as the outcome. Can be character or numeric.
baseline.Treat	Name of the reference treatment.
ID	Participant ID vector with the same length as the outcome.
response	Type of outcome. Can be "normal" for continuous outcome, "binomial" for binary outcome, "poisson" for count outcome, or "ordinal" for ordinal or nominal outcome.
model.linkfunc	Link function in the model.
model.intcpt	Form of intercept.
model.slp	Form of slope. For link function and the forms of intercept and slopes, currently implemented for 1) normal response: "identity"-"fixed/random"-"random", 2) poisson response: "log"-"fixed"-"random", 3) binomial response: "log/logit"-"fixed"-"random".
ord.ncat	Number of categories in ordinal response. The parameters for ordinal outcomes need to be tested.
ord.model	Used for ordinal outcome to pick the model. Can be "cumulative" for cumulative probability model or "acat" for adjacent category model.
ord.parallel	Whether or not the comparisons between categories or cumulative probabilities are parallel.
strata.cov	Only applicable when random intercept model is used. Stratification covariates used during randomization. Should be a data frame with the columns being the covariates and the number of rows should be equal to the length of the outcome. Must be of factor type if categorical covariates.
adjust.strata.cov	Only applicable when random intercept model is used. A vector with each element taking on possible values from c("fixed", "random") indicating how the stratification covariate should be adjusted in the model.
lvl2.cov	Participant level covariates for heterogeneous treatment effects. Should be a data frame with the columns being the covariates and the number of rows should be equal to the length of the outcome. For fixed-intercept model, participant level covariates will have interaction with treatment (slope) because there are fixed-intercepts adjusting for each participant in the model; for random-intercept model, participant level covariates will have interaction with all treatment-specific intercepts.
spline.trend	Indicator for whether the model should adjust for trend using splines. The default is F.
trend.type	"bs" for basis splines or "ns" for natural splines.
y.time	Time when the outcome is measured. Required when adjusting for trend or correlation (spline.trend/step.trend/corr.y is TRUE). Should be a vector of the same length as the outcome.
knots	Knots in y.time when adjusting for trend using splines. For trend.type = "bs", knots should be set at the end of each block except for the last block; for or trend.type = "ns", knots should be set at the end of each period except for the last period.
trend.df	Degrees of freedom for modeling splines when knots are not set.
step.trend	Indicator for whether to adjust for trend using step functions for each period.
y.step	Period number corresponding to the outcome. Should be a vector of the same length of the outcome.
corr.y	Indicator for whether the correlation among measurements shoule be modeled. The default is F.
alpha.prior	Prior for the fixed-intercepts.
beta.prior	Prior for random intercepts or slopes.
eta.prior	Prior for modelling spline terms or heterogeneous treatment effects.
dc.prior	Prior for the length between cutpoints. Used only for ordinal logistic models.
c1.prior	Prior for the first cutpoint. Used only for ordinal logistic models.
rho.prior	Prior for the correlation between random effects or the correlation among repeated measurements on each participant.
hy.prior	Prior for the variance of the residual errors for normal response, the standard deviation of random slopes for binary outcome, the variance of random effects for other types of outcomes.

jiabei-yang/nof1ins documentation built on Sept. 7, 2021, 1:10 p.m.