DynNom (version 4.0)

DynNom: Dynamic Nomograms for Linear, Generalized Linear and Proportional Hazards Models

Description

DynNom is a generic function for displaying the results of an statistical model object as a dynamic nomogram in an 'RStudio' panel or web browser. DynNom supports a variety of model objects; lm, glm, coxph and also Ols, Glm, lrm, cph models in the rms package. It is a translational tool aiming to provide easy, informative individual predictions.

Usage

DynNom(model, data, clevel = 0.95, covariate = c("slider", "numeric"), ptype = c("st", "1-st"))

Arguments

model

an lm, glm, coxph, ols, Glm, lrm or cph model object

data

dataframe containing the accompanying data

clevel

confidence level required

covariate

The option to choose the type of covariate(s) input control widget for numeric values. If "slider" (the default) is chosen a shiny application with slider control widgets are used while if "numeric" is chosen numeric values input controls will be displayed.

ptype

This plot type option relates to coxph objects only. If "st" (the default) is chosen, a plot of the estimated survivor function, S(t), is displayed. If "1-st" is chosen a plot of 1- S(t) is displayed.

Value

A dynamic nomogram in a shiny application which recognises all the predictors in the model and uses them to build a sidebar panel. It sets up drop down menus for factors and sliders set at the mean and bounded by the range for covariates.The individual predictions with a relative confidence interval are calculated using the predict function, displaying either graphically as an interactive plot in the Graphical Summary tab or a table in the Numerical Summary tab. A table of model output is also available in the Model Summary tab. In the case of the Cox proportional hazards model, estimated survivor/death function will be additionally plotted in an extra tab.

Please cite as:

Jalali, A., Roshan, D., Alvarez-Iglesias, A., Newell, J. (2016). Dynamic Nomograms for Linear, Generalized Linear and Proportional Hazard Models. R package version 3.0.

References

Banks, J. 2006. Nomograms. Encyclopedia of Statistical Sciences. 8. Easy web applications in R. http://shiny.rstudio.com Frank E Harrell Jr (2016). rms: Regression Modeling Strategies. R package version 4.5-0. https://CRAN.R-project.org/package=rms

See Also

DynNom.lm, DynNom.glm, DynNom.coxph, DynNom.ols, DynNom.lrm, DynNom.Glm, DynNom.cph

Examples

## Not run: 
# # simple linear regression models
# model1 <- lm(uptake ~ Plant + conc + Plant * conc, data = CO2)
# DynNom(model1, CO2)
# 
# data1 <- data.frame(state.x77)
# model2 <- ols(Life.Exp ~ Population + Income + Illiteracy + Murder + HS.Grad +
# Frost + Area,data=data1)
# DynNom(model2, data1)
# 
# # Generalized regression models
# data2 =as.data.frame(Titanic)
# model3 <- glm(Survived ~ Age + Class + Sex, data = data2, weights = Freq,
#               family = binomial("probit"))
# DynNom(model3, data2, clevel = 0.9)
# 
# model4 <- lrm(formula= vs ~ wt + disp, data = mtcars)
# DynNom(model4, mtcars, clevel = 0.9)
# 
# counts <- c(18, 17, 15, 20, 10, 20, 25, 13, 12)
# outcome <- gl(3, 1, 9)
# treatment <- gl(3, 3)
# data2 = data.frame(counts, outcome, treatment)
# model5 <- Glm((2 * counts) ~ outcome + treatment, family = poisson(), data = data2)
# DynNom.Glm(model5, data2)
# 
# # a proportional hazard model
# data.kidney <- kidney
# # always make sure that the categorical variables are in a factor class
# data.kidney$sex <- as.factor(data.kidney$sex)
# levels(data.kidney$sex) <- c("male", "female")
# 
# model6 <- coxph(Surv(time, status) ~ age + sex + disease, data.kidney)
# DynNom(model6, data.kidney)
# DynNom(model6, data.kidney, ptype = "1-st")
# 
# model7 <-cph((Surv(log(time), status)) ~ rcs(age, 4) * strat(trt) +
#               diagtime * strat(prior) + lsp(karno, 60), data = veteran)
# DynNom(model7, veteran)
# ## End(Not run)

if (interactive()) {
# a poisson regression model
model8 <- glm(event ~ mag + station + dist + accel, data = attenu, family = poisson)
DynNom(model8, attenu, covariate = "numeric")
}