In scientific-computing-solutions/badminton: Simulation of Continuous-Time Discrete-Space Markov Chains

Tutorial 2 -- Specifiying the transition DAG

The package allows any directed acyclic graph (DAG) to be used as the underlying transition state graph. This tutorial describes the various ways of specifying progressiong graphs to be used with the event simuator

Building the DAG from scratch

It is possible to start with an empty DAG and add nodes and edges to it:

library("badminton") #Load the badminton package
graph.par(list(nodes=list(fontsize=10))) #Set label font size

DAG <- InitializeProgressionGraph() #Empty graph
DAG <- AddNode.ProgressionGraph(DAG,nodeNames=c("N1","N2","N3","N4"))
DAG <- AddEdge.ProgressionGraph(DAG,fromNodeNames="N1",toNodeNames="N2") 
DAG <- AddEdge.ProgressionGraph(DAG,fromNodeNames=c("N2","N3"),toNodeNames=c("N3","N4")) 
plot(DAG)

Building the DAG by adding treatment arms

Often it is easier to specify the DAG in terms of treatment arms. In order to add arms, the 'death' node must first be added manually:

DAG <- InitializeProgressionGraph() #Empty graph
DAG <- AddNode.ProgressionGraph(DAG,nodeNames="death") #Add death node 

Arms can then be added:

DAG <- AddArm.ProgressionGraph(DAG,armName="c",
                                   armProgression=c("s1","s2","s3"),deathNode="death",edges="immediate")
plot(DAG)

The nodes added are armName.armProgression[1], armName.armProgression[2],... The arm ends at the deathNode node (which must already exist). There are three possible edges options: "immediate": Subjects must progress sequentially through all states of the arm (see above) "death": Subjects must progress sequentially through all states of the arm or progress straight to the deathNode

DAG <- AddArm.ProgressionGraph(DAG,armName="a",
                                   armProgression=c("s1","s2","s3","s4"),deathNode="death",edges="death")
plot(DAG)

• "all" : Subjects can jump forward to any later state in this arm

DAG <- AddArm.ProgressionGraph(DAG,armName="a2",
                                   armProgression=c("s1","s2","s3"),deathNode="death",edges="all")
plot(DAG)

It is possible to add further nodes and edges using the function from earlier in this tutorial

Using SimpleStudyProgressionGraph

Often, all arms have the same progression sequence and there is a set of extra edges denoting treatment crossovers. If this is the case then the SimpleStudyProgressionGraph function can be used. This function creates a graph containing the node death and adds a set of arms. The arms argument contains a vector of arms, the armProgression argument contains a vector of progressions of these arms, the edge option is the same as the AddArm function above and crossOver contains a set of additional crossover edges to be added into the graph.

crossOver <- c("c.s1","a.s2","a.s2","a2.s3")
DAG <-  SimpleStudyProgressionGraph(arms=c("c","a","a2"),
                                 armProgression=c("s1","s2","s3"),edge="immediate",crossOver=crossOver)
plot(DAG)

Note: The crossOver edges have their isResetEdge property set to TRUE, see Tutorial 3 for further details

scientific-computing-solutions/badminton documentation built on May 29, 2019, 3:43 p.m.