optimize_designs: Adaptive Enrichment Design Optimization Using Simulated...
In mrosenblum/AdaptiveDesignOptimizer: Optimizes Adaptive Enrichment Designs
Description
Usage
Arguments
Value
Designs
design.parameters
design.performance
Examples
View source: R/optimize_designs.R
Description
Adaptive Enrichment Design Optimization Using Simulated Annealing
Authors: Josh Betz (jbetz@jhu.edu), Tianchen Qian, Michael Rosenblum
Usage
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
optimize_designs(ui.n.arms, ui.type.of.outcome.data,
ui.time.to.event.trial.type,
ui.time.to.event.non.inferiority.trial.margin, ui.subpopulation.1.size,
ui.total.alpha, ui.max.size, ui.max.duration, ui.accrual.yearly.rate,
ui.followup.length, ui.optimization.target,
ui.time.to.event.censoring.rate, ui.mcid, ui.incorporate.precision.gain,
ui.relative.efficiency, ui.max.stages, ui.include.designs.start.subpop.1,
ui.population.parameters, ui.desired.power, ui.scenario.weights,
min.n.per.arm = 25, min.enrollment.period = 0.5,
simulated.annealing.parameter.function.scale = 1,
simulated.annealing.parameter.n.scale = 100,
simulated.annealing.parameter.period.scale = 2,
simulated.annealing.parameter.max.iterations = 1000,
simulated.annealing.parameter.n.simulations = 10000,
simulated.annealing.parameter.means.temperature = 100,
simulated.annealing.parameter.survival.temperature = 10,
simulated.annealing.parameter.evals.per.temp = 10,
simulated.annealing.parameter.report.iteration = 1,
simulated.annealing.parameter.power.penalty = 1e+05)
Arguments
ui.n.arms
(the prefix 'ui' abbreviates 'user-input') Number of Arms (including control arm), e.g., 2 arms means one treatment arm and one control arm; the design classes that come with this package can handle 2 or 3 arms
ui.type.of.outcome.data
"continuous", "binary", or "time-to-event" (i.e., survival outcome)
ui.time.to.event.trial.type
"superiority" or "non-inferiority" trial type; only implemented for time-to-event outcomes
ui.time.to.event.non.inferiority.trial.margin
Non-inferiority margin; only relevant if outcome is time-to-event and trial type is non-inferiority. Represented as hazard ratio, required to be at least 1.
ui.subpopulation.1.size
Proportion of overall population in subpopulation 1. Must be between 0 and 1.
ui.total.alpha
Familywise Type I error rate (1-sided)
ui.max.size
Maximum allowed sample size
ui.max.duration
Maximum allowed trial duration in years
ui.accrual.yearly.rate
Number of participants enrolled per year; assumed constant throughout trial
ui.followup.length
Time from enrollment to measurement of primary outcome (only used for continuous or binary outcome types)
ui.optimization.target
Quantity being optimized (objective function); "size" represents expected sample size
ui.time.to.event.censoring.rate
probability that primary outcome is censored, assumed to be independent of the outcome and subpopulation; only implemented for time-to-event outcomes
ui.mcid
Minimum, Clinically Important Treatment Effect (as difference of population means for binary/continuous outcomes; as hazard ratio for time-to-event outcomes)
ui.incorporate.precision.gain
Incorporate into analysis a precision gain from adjustment for prognostic baseline variables; allowed values: TRUE or FALSE
ui.relative.efficiency
If ui.incorporate.precision.gain==TRUE, this specifies relative efficiency (number > 1), representing the assumed precision gain from adjustment for prognostic baseline variables.
ui.max.stages
Maximum number of stages allowed (currently not used by default classes of trial designs)
ui.include.designs.start.subpop.1
Search over designs that allow only subpopulation 1 to be enrolled during stage 1; TRUE or FALSE
ui.population.parameters
Matrix encoding scenarios (data generating distributions) used to define power constraints and objective function
ui.desired.power
Matrix encoding power requirements for each scenario
ui.scenario.weights
Matrix encoding weights used to define objective function
min.n.per.arm
Minimum sample size per arm allowed
min.enrollment.period
Minimum enrollment duration for trial
simulated.annealing.parameter.function.scale
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.n.scale
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.period.scale
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.max.iterations
Maximum Number of Different Designs to Search Over using Simluated Annealing optimization
simulated.annealing.parameter.n.simulations
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.means.temperature
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.survival.temperature
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.evals.per.temp
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.report.iteration
Used by Simulated Annealing Optimization algorithm
simulated.annealing.parameter.power.penalty
Used in Objective Function to incorporate Power Constraints by Simulated Annealing Optimization algorithm
Value
4 element list containing optimized designs from four classes (with increasing complexity):
Designs
(first two not adaptive; last two adaptive)
Single.Stage.Equal.Alpha.Allocation.Design
Single.Stage.Optimized.Alpha.Allocation.Design
Two.Stage.Equal.Alpha.Allocation.Design
Two.Stage.Optimized.Alpha.Allocation.Design
Each optimized design is a list containing: design.parameters and design.performance
design.parameters
design.parameters has the following elements:
cumulative.sample.sizes.and.calendar.time.per.stage The cumulative number enrolled (if no early stopping) per stage and calendar times of analyses just after each stage. In column names, “A” and “C” denote the treatment arm and control arm, respectively; numbers 1 and 2 indicate the corresponding subpopulation. Sample sizes represent the number enrolled at the time of the corresponding analysis (which may exceed the number of participants with outcomes observed, due to the time between enrollment and outcome measurement for each participant
alpha.allocation=Alpha allocation using Error Spending Approach
futility.boundaries=Boundaries for stopping subpopulation accrual, on the z-scale (or in designs with more than one treatment arm compared to control, this is gives for each treatment arm by subpopulation combination.
design.performance
design.performance contains the following values:
Power=Power to reject each null hypothesis under each scenario (NA indicates null hypothesis is true, so no power is presented)
Type.1.Error=Type I error for each null hypothesis under each scenario (NA indicates null hypothesis is false)
Expected.Sample.Size
Expected.Duration (in years)
Distribution.of.sample.size.and.duration.per.scenario=For each scenario, every possible combination of early stopping is considered. Columns C1, A1, etc. have the same meaning as described about for sample.sizes.and.calendar.time.per.stage. The value listed under each such column gives the analysis number at which accrual for that arm by subpopulation combination is stopped. E.g., C1=2,C2=1,A1=2,A2=1 corresponds to stopping the control and treatment A for subpopualtion 1 at the end of stage 1, while these continue to the end of stage 2 for subpopulation 2. The subsequent columns give the sample size, duration, and person-time when this pattern occurs. The columns frequency and proportion tell how often this pattern occurred under the corresponding scenario number (based on simulation).
Examples
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
#For demonstration purposes, the examples below only execute 2 iterations of simulated annealing.
#In general, it is recommended to use at least 500 iterations.
#Example 1: Time-to-event outcome; 1 treatment arm versus control; non-inferiority design
optimized_designs <- optimize_designs(
ui.n.arms=2,
ui.type.of.outcome.data="time-to-event",
ui.time.to.event.trial.type="non-inferiority",
ui.time.to.event.non.inferiority.trial.margin=1.35,
ui.subpopulation.1.size=0.33,
ui.total.alpha=0.05,
ui.max.size=10000,
ui.max.duration=10,
ui.accrual.yearly.rate=1000,
ui.followup.length=0,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=0.1,
ui.incorporate.precision.gain=FALSE,
ui.relative.efficiency=1,
ui.max.stages=2,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters= 0.08*matrix(c(1.00, 1.00, 1.00, 1.00, 1.00, 1.00, 1.00, 1.3500001,
1.00, 1.00, 1.00, 2.14, 1.00, 1.00, 1.3500001, 1.3500001), ncol=4, byrow=TRUE,
dimnames=list(c(),c("lambda1_con","lambda2_con","lambda1_trt","lambda2_trt"))),
ui.desired.power=0.8*matrix(c(1.00, 1.00, 0, 1.00, 0, 0, 1.00, 0, 0, 0, 0, 0), ncol=3,
byrow=TRUE,dimnames=list(c(),c("Pow_H(0,1)","Pow_H(0,2)","Pow_Reject_H0,1_and_H0,2"))),
ui.scenario.weights=matrix(rep(0.25,4),ncol=1,dimnames=list(c(),c("weight"))),
simulated.annealing.parameter.max.iterations=2
)
#Example 2: continuous outcome; 1 treatment arm versus control; superiority design
optimized_designs <- optimize_designs(
ui.n.arms=2,
ui.type.of.outcome.data="continuous",
ui.time.to.event.trial.type="",
ui.time.to.event.non.inferiority.trial.margin=NULL,
ui.subpopulation.1.size=0.5,
ui.total.alpha=0.05,
ui.max.size=1000,
ui.max.duration=5,
ui.accrual.yearly.rate=250,
ui.followup.length=1,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=NULL,
ui.incorporate.precision.gain=FALSE,
ui.relative.efficiency=1,
ui.max.stages=5,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters=matrix(c(15,15,3600,3600,3600,3600,15,0,3600,3600,3600,3600,
0,15,3600,3600,3600,3600,0,0,3600,3600,3600,3600),nrow=4, ncol=6, byrow=TRUE,dimnames=
list(c(),c("delta1","delta2","sigma1_trt","sigma1_con","sigma2_trt","sigma2_con"))),
ui.desired.power=matrix(c(0,0,0.8,0.8,0,0,0,0.8,0,0,0,0), nrow=4, ncol=3, byrow=TRUE,
dimnames=list(c(),c("Pow_H(0,1)","Pow_H(0,2)","Pow_Reject_H0,1_and_H0,2"))),
ui.scenario.weights=matrix(c(0.25,0.25,0.25,0.25),ncol=1,dimnames=list(c(),c("weight"))),
simulated.annealing.parameter.max.iterations=2
)
#Example 3: binary outcome; 1 treatment arm versus control; superiority design
optimized_designs <- optimize_designs(
ui.n.arms=2,
ui.type.of.outcome.data="binary",
ui.time.to.event.trial.type="",
ui.time.to.event.non.inferiority.trial.margin=NULL,
ui.subpopulation.1.size=0.4,
ui.total.alpha=0.05,
ui.max.size=2000,
ui.max.duration=5,
ui.accrual.yearly.rate=400,
ui.followup.length=0,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=NULL,
ui.incorporate.precision.gain=TRUE,
ui.relative.efficiency=1.2,
ui.max.stages=5,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters=matrix(c(0.4,0.3,0.5,0.4,0.4,0.3,0.4,0.4,0.3,0.3,0.4,0.4),
nrow=3, ncol=4, byrow=TRUE,dimnames=list(c(),c("p1_trt","p1_con","p2_trt","p2_con"))),
ui.desired.power=matrix(c(0,0,0.8,0.8,0,0,0,0,0), nrow=3, ncol=3, byrow=TRUE,
dimnames=list(c(),c("Pow_H(0,1)","Pow_H(0,2)","Pow_Reject_H0,1_and_H0,2"))),
ui.scenario.weights=matrix(c(0.33,0.33,0.34),ncol=1,dimnames=list(c(),c("weight"))),
simulated.annealing.parameter.max.iterations=2
)
#Example 4: continuous outcome; 2 treatment arms versus control; superiority design
optimized_designs <- optimize_designs(
ui.n.arms=3,
ui.type.of.outcome.data="continuous",
ui.time.to.event.trial.type="",
ui.time.to.event.non.inferiority.trial.margin=NULL,
ui.subpopulation.1.size=0.49,
ui.total.alpha=0.05,
ui.max.size=3000,
ui.max.duration=8,
ui.accrual.yearly.rate=240,
ui.followup.length=0.5,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=NULL,
ui.incorporate.precision.gain=TRUE,
ui.relative.efficiency=1,
ui.max.stages=4,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters=matrix(c(0,3600,0,3600,0,3600,0,3600,0,3600,0,3600,0,3600,0,3600,
15,3600,0,3600,0,3600,0,3600,0,3600,0,3600,15,3600,0,3600,15,3600,0,3600,0,3600,0,3600,15,
3600,15,3600,0,3600,0,3600,0,3600,0,3600,15,3600,15,3600,15,3600,0,3600,0,3600,0,3600,15,
3600,15,3600,15,3600,15,3600), nrow=6, ncol=12, byrow=TRUE),
ui.desired.power=matrix(c(
0,0,0,0,0,
0.8,0,0,0,0,
0.8,0,0.8,0,0,
0.8,0.8,0,0,0,
0.8,0.8,0.8,0,0,
0.8,0.8,0.8,0.8,0),
nrow=6, ncol=5, byrow=TRUE),
ui.scenario.weights=matrix(c(0.166,0.166,0.166,0.166,0.166,0.167)),
simulated.annealing.parameter.max.iterations=2)
mrosenblum/AdaptiveDesignOptimizer documentation built on July 10, 2019, 2:46 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Description Usage Arguments Value Designs design.parameters design.performance Examples
View source: R/optimize_designs.R
Description
Adaptive Enrichment Design Optimization Using Simulated Annealing Authors: Josh Betz (jbetz@jhu.edu), Tianchen Qian, Michael Rosenblum
Usage
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 | optimize_designs(ui.n.arms, ui.type.of.outcome.data,
ui.time.to.event.trial.type,
ui.time.to.event.non.inferiority.trial.margin, ui.subpopulation.1.size,
ui.total.alpha, ui.max.size, ui.max.duration, ui.accrual.yearly.rate,
ui.followup.length, ui.optimization.target,
ui.time.to.event.censoring.rate, ui.mcid, ui.incorporate.precision.gain,
ui.relative.efficiency, ui.max.stages, ui.include.designs.start.subpop.1,
ui.population.parameters, ui.desired.power, ui.scenario.weights,
min.n.per.arm = 25, min.enrollment.period = 0.5,
simulated.annealing.parameter.function.scale = 1,
simulated.annealing.parameter.n.scale = 100,
simulated.annealing.parameter.period.scale = 2,
simulated.annealing.parameter.max.iterations = 1000,
simulated.annealing.parameter.n.simulations = 10000,
simulated.annealing.parameter.means.temperature = 100,
simulated.annealing.parameter.survival.temperature = 10,
simulated.annealing.parameter.evals.per.temp = 10,
simulated.annealing.parameter.report.iteration = 1,
simulated.annealing.parameter.power.penalty = 1e+05)
|
Arguments
ui.n.arms |
(the prefix 'ui' abbreviates 'user-input') Number of Arms (including control arm), e.g., 2 arms means one treatment arm and one control arm; the design classes that come with this package can handle 2 or 3 arms |
ui.type.of.outcome.data |
"continuous", "binary", or "time-to-event" (i.e., survival outcome) |
ui.time.to.event.trial.type |
"superiority" or "non-inferiority" trial type; only implemented for time-to-event outcomes |
ui.time.to.event.non.inferiority.trial.margin |
Non-inferiority margin; only relevant if outcome is time-to-event and trial type is non-inferiority. Represented as hazard ratio, required to be at least 1. |
ui.subpopulation.1.size |
Proportion of overall population in subpopulation 1. Must be between 0 and 1. |
ui.total.alpha |
Familywise Type I error rate (1-sided) |
ui.max.size |
Maximum allowed sample size |
ui.max.duration |
Maximum allowed trial duration in years |
ui.accrual.yearly.rate |
Number of participants enrolled per year; assumed constant throughout trial |
ui.followup.length |
Time from enrollment to measurement of primary outcome (only used for continuous or binary outcome types) |
ui.optimization.target |
Quantity being optimized (objective function); "size" represents expected sample size |
ui.time.to.event.censoring.rate |
probability that primary outcome is censored, assumed to be independent of the outcome and subpopulation; only implemented for time-to-event outcomes |
ui.mcid |
Minimum, Clinically Important Treatment Effect (as difference of population means for binary/continuous outcomes; as hazard ratio for time-to-event outcomes) |
ui.incorporate.precision.gain |
Incorporate into analysis a precision gain from adjustment for prognostic baseline variables; allowed values: TRUE or FALSE |
ui.relative.efficiency |
If ui.incorporate.precision.gain==TRUE, this specifies relative efficiency (number > 1), representing the assumed precision gain from adjustment for prognostic baseline variables. |
ui.max.stages |
Maximum number of stages allowed (currently not used by default classes of trial designs) |
ui.include.designs.start.subpop.1 |
Search over designs that allow only subpopulation 1 to be enrolled during stage 1; TRUE or FALSE |
ui.population.parameters |
Matrix encoding scenarios (data generating distributions) used to define power constraints and objective function |
ui.desired.power |
Matrix encoding power requirements for each scenario |
ui.scenario.weights |
Matrix encoding weights used to define objective function |
min.n.per.arm |
Minimum sample size per arm allowed |
min.enrollment.period |
Minimum enrollment duration for trial |
simulated.annealing.parameter.function.scale |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.n.scale |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.period.scale |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.max.iterations |
Maximum Number of Different Designs to Search Over using Simluated Annealing optimization |
simulated.annealing.parameter.n.simulations |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.means.temperature |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.survival.temperature |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.evals.per.temp |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.report.iteration |
Used by Simulated Annealing Optimization algorithm |
simulated.annealing.parameter.power.penalty |
Used in Objective Function to incorporate Power Constraints by Simulated Annealing Optimization algorithm |
Value
4 element list containing optimized designs from four classes (with increasing complexity):
Designs
(first two not adaptive; last two adaptive) Single.Stage.Equal.Alpha.Allocation.Design
Single.Stage.Optimized.Alpha.Allocation.Design
Two.Stage.Equal.Alpha.Allocation.Design
Two.Stage.Optimized.Alpha.Allocation.Design
Each optimized design is a list containing: design.parameters and design.performance
design.parameters
design.parameters has the following elements: cumulative.sample.sizes.and.calendar.time.per.stage The cumulative number enrolled (if no early stopping) per stage and calendar times of analyses just after each stage. In column names, “A” and “C” denote the treatment arm and control arm, respectively; numbers 1 and 2 indicate the corresponding subpopulation. Sample sizes represent the number enrolled at the time of the corresponding analysis (which may exceed the number of participants with outcomes observed, due to the time between enrollment and outcome measurement for each participant
alpha.allocation=Alpha allocation using Error Spending Approach
futility.boundaries=Boundaries for stopping subpopulation accrual, on the z-scale (or in designs with more than one treatment arm compared to control, this is gives for each treatment arm by subpopulation combination.
design.performance
design.performance contains the following values: Power=Power to reject each null hypothesis under each scenario (NA indicates null hypothesis is true, so no power is presented)
Type.1.Error=Type I error for each null hypothesis under each scenario (NA indicates null hypothesis is false)
Expected.Sample.Size
Expected.Duration (in years)
Distribution.of.sample.size.and.duration.per.scenario=For each scenario, every possible combination of early stopping is considered. Columns C1, A1, etc. have the same meaning as described about for sample.sizes.and.calendar.time.per.stage. The value listed under each such column gives the analysis number at which accrual for that arm by subpopulation combination is stopped. E.g., C1=2,C2=1,A1=2,A2=1 corresponds to stopping the control and treatment A for subpopualtion 1 at the end of stage 1, while these continue to the end of stage 2 for subpopulation 2. The subsequent columns give the sample size, duration, and person-time when this pattern occurs. The columns frequency and proportion tell how often this pattern occurred under the corresponding scenario number (based on simulation).
Examples
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 | #For demonstration purposes, the examples below only execute 2 iterations of simulated annealing.
#In general, it is recommended to use at least 500 iterations.
#Example 1: Time-to-event outcome; 1 treatment arm versus control; non-inferiority design
optimized_designs <- optimize_designs(
ui.n.arms=2,
ui.type.of.outcome.data="time-to-event",
ui.time.to.event.trial.type="non-inferiority",
ui.time.to.event.non.inferiority.trial.margin=1.35,
ui.subpopulation.1.size=0.33,
ui.total.alpha=0.05,
ui.max.size=10000,
ui.max.duration=10,
ui.accrual.yearly.rate=1000,
ui.followup.length=0,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=0.1,
ui.incorporate.precision.gain=FALSE,
ui.relative.efficiency=1,
ui.max.stages=2,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters= 0.08*matrix(c(1.00, 1.00, 1.00, 1.00, 1.00, 1.00, 1.00, 1.3500001,
1.00, 1.00, 1.00, 2.14, 1.00, 1.00, 1.3500001, 1.3500001), ncol=4, byrow=TRUE,
dimnames=list(c(),c("lambda1_con","lambda2_con","lambda1_trt","lambda2_trt"))),
ui.desired.power=0.8*matrix(c(1.00, 1.00, 0, 1.00, 0, 0, 1.00, 0, 0, 0, 0, 0), ncol=3,
byrow=TRUE,dimnames=list(c(),c("Pow_H(0,1)","Pow_H(0,2)","Pow_Reject_H0,1_and_H0,2"))),
ui.scenario.weights=matrix(rep(0.25,4),ncol=1,dimnames=list(c(),c("weight"))),
simulated.annealing.parameter.max.iterations=2
)
#Example 2: continuous outcome; 1 treatment arm versus control; superiority design
optimized_designs <- optimize_designs(
ui.n.arms=2,
ui.type.of.outcome.data="continuous",
ui.time.to.event.trial.type="",
ui.time.to.event.non.inferiority.trial.margin=NULL,
ui.subpopulation.1.size=0.5,
ui.total.alpha=0.05,
ui.max.size=1000,
ui.max.duration=5,
ui.accrual.yearly.rate=250,
ui.followup.length=1,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=NULL,
ui.incorporate.precision.gain=FALSE,
ui.relative.efficiency=1,
ui.max.stages=5,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters=matrix(c(15,15,3600,3600,3600,3600,15,0,3600,3600,3600,3600,
0,15,3600,3600,3600,3600,0,0,3600,3600,3600,3600),nrow=4, ncol=6, byrow=TRUE,dimnames=
list(c(),c("delta1","delta2","sigma1_trt","sigma1_con","sigma2_trt","sigma2_con"))),
ui.desired.power=matrix(c(0,0,0.8,0.8,0,0,0,0.8,0,0,0,0), nrow=4, ncol=3, byrow=TRUE,
dimnames=list(c(),c("Pow_H(0,1)","Pow_H(0,2)","Pow_Reject_H0,1_and_H0,2"))),
ui.scenario.weights=matrix(c(0.25,0.25,0.25,0.25),ncol=1,dimnames=list(c(),c("weight"))),
simulated.annealing.parameter.max.iterations=2
)
#Example 3: binary outcome; 1 treatment arm versus control; superiority design
optimized_designs <- optimize_designs(
ui.n.arms=2,
ui.type.of.outcome.data="binary",
ui.time.to.event.trial.type="",
ui.time.to.event.non.inferiority.trial.margin=NULL,
ui.subpopulation.1.size=0.4,
ui.total.alpha=0.05,
ui.max.size=2000,
ui.max.duration=5,
ui.accrual.yearly.rate=400,
ui.followup.length=0,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=NULL,
ui.incorporate.precision.gain=TRUE,
ui.relative.efficiency=1.2,
ui.max.stages=5,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters=matrix(c(0.4,0.3,0.5,0.4,0.4,0.3,0.4,0.4,0.3,0.3,0.4,0.4),
nrow=3, ncol=4, byrow=TRUE,dimnames=list(c(),c("p1_trt","p1_con","p2_trt","p2_con"))),
ui.desired.power=matrix(c(0,0,0.8,0.8,0,0,0,0,0), nrow=3, ncol=3, byrow=TRUE,
dimnames=list(c(),c("Pow_H(0,1)","Pow_H(0,2)","Pow_Reject_H0,1_and_H0,2"))),
ui.scenario.weights=matrix(c(0.33,0.33,0.34),ncol=1,dimnames=list(c(),c("weight"))),
simulated.annealing.parameter.max.iterations=2
)
#Example 4: continuous outcome; 2 treatment arms versus control; superiority design
optimized_designs <- optimize_designs(
ui.n.arms=3,
ui.type.of.outcome.data="continuous",
ui.time.to.event.trial.type="",
ui.time.to.event.non.inferiority.trial.margin=NULL,
ui.subpopulation.1.size=0.49,
ui.total.alpha=0.05,
ui.max.size=3000,
ui.max.duration=8,
ui.accrual.yearly.rate=240,
ui.followup.length=0.5,
ui.optimization.target="size",
ui.time.to.event.censoring.rate=0,
ui.mcid=NULL,
ui.incorporate.precision.gain=TRUE,
ui.relative.efficiency=1,
ui.max.stages=4,
ui.include.designs.start.subpop.1=FALSE,
ui.population.parameters=matrix(c(0,3600,0,3600,0,3600,0,3600,0,3600,0,3600,0,3600,0,3600,
15,3600,0,3600,0,3600,0,3600,0,3600,0,3600,15,3600,0,3600,15,3600,0,3600,0,3600,0,3600,15,
3600,15,3600,0,3600,0,3600,0,3600,0,3600,15,3600,15,3600,15,3600,0,3600,0,3600,0,3600,15,
3600,15,3600,15,3600,15,3600), nrow=6, ncol=12, byrow=TRUE),
ui.desired.power=matrix(c(
0,0,0,0,0,
0.8,0,0,0,0,
0.8,0,0.8,0,0,
0.8,0.8,0,0,0,
0.8,0.8,0.8,0,0,
0.8,0.8,0.8,0.8,0),
nrow=6, ncol=5, byrow=TRUE),
ui.scenario.weights=matrix(c(0.166,0.166,0.166,0.166,0.166,0.167)),
simulated.annealing.parameter.max.iterations=2)
|
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.