Select treatments to advance based on a Bayesian rule to select any that have at least a user-specified probability of being greater than a user-specified historical response rate.
SelectExpUsingBayesianRule.RdThis function is used for the MAMS design with a binary outcome and will perform treatment selection at the interim analysis (IA). At the IA, utilize a Bayesian rule to select any experimental treatment that has at least a user-specified probability (treatmentPValue) of being greater than a user-specified historical response rate (historicResponseRate). Specifically, if Pr( pj > historicResponseRate | data ) > treatmentPValue, then experimental treatment j is selected for stage 2. If none of the treatments meet the criteria for selection, then select the treatment with the largest Pr( pj > historiceResponseRate | data ). User-specified pj ~ Beta( dPriorAlpha, dPriorBeta ) After the IA, use a randomization ratio of 2:1 (experimental:control) for all experimental treatments that are selected for stage 2.
Arguments
- SimData
Data frame which consists of data generated in current simulation
- DesignParam
List of Design and Simulation Parameters required to perform treatment selection.
- LookInfo
List containing Design and Simulation Parameters, which might be required to perform treatment selection
- UserParam
A list of user defined parameters in East. The default must be NULL. If UserParam is supplied, the list must contain the following named element:
- UserParam$dPriorAlpha
A value (0,1) that defines the prior alpha parameter of the beta distribution. If this value is not specified, the default is 0.2.
- UserParam$dPriorBeta
A value (0,1) that specifies the prior beta parameter of the beta distribution. If this value is not specified, the default is 0.8.
- UserParam$historicResponseRate
A value (0,1) that specifies the historic response rate. If this value is not specified, the default is 0.1.
- UserParam$treatmentPValue
A value (0,1) that specifies the probability needed of being greater than the historic response rate for an experimental treatment to be selected. If this value is not specified, the default is 0.2.
Value
TreatmentID A vector that consists of the experimental treatments that were selected and carried forward. Experimental treatment IDs are 1, 2, ..., number of experimental treatments
AllocRatio A vector that consists of the allocation for all experimental treatments that continue to the next phase.
ErrorCode An integer value: ErrorCode = 0 --> No Error ErrorCode > 0 --> Non fatal error, current simulation is aborted but the next simulations will run ErrorCode < 0 --> Fatal error, no further simulation will be attempted
Note
The length of TreatmentID and AllocRatio must be the same.
The allocation ratio for control will be 1, AllocRatio are relative to this value. So, a 2 will randomize twice as many to experimental
The order of AllocRatio should be the same as TreatmentID, and the corresponding elements will have the assigned allocation ratio
The returned vector ONLY includes TreatmentIDs for experimental treatments, eg TreatmentID = c( 0, 1, 2 ) is invalid, because you do NOT need to include 0 for control.
You must return at LEAST one treatment and one allocation ratio
Helpful Hints: There is often info that East sends to R that are not shown in a given example. It can be very helpful to save the input objects and then load them into your R session and inspect them. This can be done with the following R code in your function.
saveRDS( SimData, "SimData.Rds") saveRDS( DesignParam, "DesignParam.Rds" ) saveRDS( LookInfo, "LookInfo.Rds" )
The above code will save each of the input objects to a file so they may be examined within R.