### Install clinfun package

Source: https://context7.com/veseshan/clinfun/llms.txt

Instructions for installing the package from CRAN or the development version from GitHub.

```r
# Install from CRAN
install.packages("clinfun")

# Or install development version from GitHub
# install.packages("devtools")
devtools::install_github("veseshan/clinfun")
```

--------------------------------

### Install clinfun from GitHub

Source: https://github.com/veseshan/clinfun/blob/main/Readme.md

Use this command to install the development version of the package directly from GitHub using devtools.

```r
# install.packages("devtools")
devtools::install_github("veseshan/clinfun")
```

--------------------------------

### Install clinfun from CRAN

Source: https://github.com/veseshan/clinfun/blob/main/Readme.md

Use this command to install the stable version of the package from the CRAN repository.

```r
install.packages("clinfun")
```

--------------------------------

### Load clinfun library

Source: https://context7.com/veseshan/clinfun/llms.txt

Load the package into the R environment.

```r
library(clinfun)
```

--------------------------------

### Create Futility Monitoring Stopping Rules

Source: https://context7.com/veseshan/clinfun/llms.txt

Creates stopping boundaries for futility monitoring in single-stage Phase II designs using futilbdry.

```r
library(clinfun)

# Create futility stopping rule for a single-stage design:
# - Stop if response rate appears below 25% (rLo)
# - Target response rate is 45% (rHi)
# - Monitor at every 5 patients up to 50
futility_rule <- futilbdry(
  rLo = 0.25,                    # null response rate
  rHi = 0.45,                    # alternative response rate
  n = seq(5, 50, by = 5),        # monitoring schedule
  size = 0.1,                    # Type I error
  power = 0.9                    # power requirement
)

print(futility_rule)
```

--------------------------------

### ph2simon - Simon's Two-Stage Phase II Design

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates optimal and minimax two-stage Phase II designs using Simon's methodology.

```APIDOC
## ph2simon

### Description
Calculates optimal and minimax two-stage Phase II designs using Simon's methodology. The function determines sample sizes and response thresholds that minimize either expected sample size (optimal) or maximum sample size (minimax) while controlling Type I and Type II error rates.

### Parameters
#### Arguments
- **pu** (numeric) - Required - Unacceptable response rate
- **pa** (numeric) - Required - Desirable response rate
- **ep1** (numeric) - Required - Type I error rate
- **ep2** (numeric) - Required - Type II error rate

### Request Example
ph2simon(pu = 0.05, pa = 0.2, ep1 = 0.1, ep2 = 0.1)

### Response
#### Success Response (200)
- **r1** (integer) - Stage 1 response threshold
- **n1** (integer) - Stage 1 sample size
- **r** (integer) - Overall response threshold
- **n** (integer) - Total sample size
- **EN(p0)** (numeric) - Expected sample size under null
- **PET(p0)** (numeric) - Probability of early termination under null
```

--------------------------------

### Calculate Simon's Two-Stage Phase II Design

Source: https://context7.com/veseshan/clinfun/llms.txt

Determine sample sizes and response thresholds for two-stage trials using Simon's methodology.

```r
library(clinfun)

# Design a two-stage trial where:
# - 5% response rate is unacceptable (pu)
# - 20% response rate is desirable (pa)
# - Type I error (ep1) = 10%
# - Type II error (ep2) = 10%
trial <- ph2simon(pu = 0.05, pa = 0.2, ep1 = 0.1, ep2 = 0.1)

# Print the optimal and minimax designs
print(trial)
#>  Simon 2-stage  Phase II design
#>
#> Unacceptable response rate:  0.05
#> Desirable response rate:  0.2
#> Error rates: alpha =  0.1 ; beta =  0.1
#>
#>          r1 n1  r  n EN(p0) PET(p0)
#> Optimal   0 13  3 37 22.72   0.513
#> Minimax   0 18  3 32 25.51   0.397

# Plot the design space showing optimal (O) and minimax (M) designs
plot(trial)

# Get admissible designs (compromise between optimal and minimax)
twostage.admissible(trial)
#>         r1 n1 r  n    EN   PET  qLo  qHi
#> Minimax  0 18 3 32 25.51 0.397 0.00 0.08
#>          0 15 3 33 24.27 0.463 0.08 0.17
#>          0 14 3 35 23.64 0.488 0.17 0.57
#> Optimal  0 13 3 37 22.72 0.513 0.57 1.00
```

--------------------------------

### Calculate sample sizes with fe.ssize

Source: https://github.com/veseshan/clinfun/blob/main/Readme.md

Calculates CPS and Fisher's exact sample sizes with power estimates for specified response rates.

```r
library(clinfun)

fe.ssize(p1 = .8,
         p2 = .2,
         alpha=0.05,
         power=0.8)
```

--------------------------------

### Compute Toxicity Monitoring Stopping Rules

Source: https://context7.com/veseshan/clinfun/llms.txt

Computes continuous stopping boundaries for toxicity monitoring using toxbdry.

```r
library(clinfun)

# Create toxicity stopping rule:
# - Stop if toxicity rate appears to exceed 35% (pHi)
# - Acceptable toxicity rate is 20% (pLo)
# - Monitor at 20, 40, 60, and 75 patients
tox_rule <- toxbdry(
  pLo = 0.2,              # acceptable toxicity rate
  pHi = 0.35,             # unacceptable toxicity rate
  n = c(20, 40, 60, 75),  # monitoring times
  cP0 = 0.1,              # Type I error (false positive)
  cP1 = 0.9               # Power to detect high toxicity
)

print(tox_rule)

# Continuous monitoring example (every patient from 2 to 30)
continuous_rule <- toxbdry(pLo = 0.1, pHi = 0.3, n = 2:30)
print(continuous_rule)
```

--------------------------------

### Perform Statistical Inference After Two-Stage Trial

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculate p-values, UMVUE estimates, and confidence intervals for completed two-stage trials.

```r
library(clinfun)

# After completing a two-stage trial with:
# - r1=0, n1=18 (stage 1: stop if 0 responses in 18 patients)
# - r=3, n=32 (overall: recommend if >3 responses in 32 patients)
# Observed: x=5 responses total
result <- twostage.inference(
  x = 5,      # observed responses
  r1 = 0,     # stage 1 threshold
  n1 = 18,    # stage 1 sample size
  n = 32,     # total sample size
  pu = 0.05,  # null hypothesis response rate
  alpha = 0.1
)

print(result)
#> UMVUE:     0.1563
#> p-value:   0.0146
#> 80% CI:    (0.0741, 0.2615)
```

--------------------------------

### Calculate Single-Stage Phase II Design

Source: https://context7.com/veseshan/clinfun/llms.txt

Compute exact single-stage designs for scenarios where two-stage designs are impractical.

```r
library(clinfun)

# Design a single-stage trial where:
# - 40% response rate is unacceptable
# - 60% response rate is desirable
# - Type I error = 10%
# - Type II error = 10%
design <- ph2single(pu = 0.4, pa = 0.6, ep1 = 0.1, ep2 = 0.1)

print(design)
#>    n  r   Type I    Type II
#> 1 41 20 0.097449 0.09701686
#> 2 46 23 0.094047 0.08432963
#> 3 51 26 0.092067 0.07477155
#> 4 56 29 0.091147 0.06730206
#> 5 61 32 0.091016 0.06129671

# Interpretation: Enroll n=41 patients, treatment is recommended if >20 responses
```

--------------------------------

### futilbdry - Futility Monitoring Stopping Rule

Source: https://context7.com/veseshan/clinfun/llms.txt

Creates stopping boundaries for futility monitoring in single-stage Phase II designs.

```APIDOC
## futilbdry

### Description
Creates stopping boundaries for futility monitoring in single-stage Phase II designs.

### Method
`futilbdry(rLo, rHi, n, size, power)`

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
* **rLo** (numeric) - Null response rate (lower bound for efficacy).
* **rHi** (numeric) - Alternative response rate (target efficacy).
* **n** (numeric vector) - Sample sizes at which to monitor.
* **size** (numeric) - Type I error rate (alpha).
* **power** (numeric) - Desired power (1 - beta).

### Request Example
```r
library(clinfun)
futility_rule <- futilbdry(
  rLo = 0.25,
  rHi = 0.45,
  n = seq(5, 50, by = 5),
  size = 0.1,
  power = 0.9
)
print(futility_rule)
```

### Response
#### Success Response (200)
Returns a data frame with monitoring schedule, futility boundary, and probabilities of stopping or effective rates at each monitoring point.

#### Response Example
```
Futility boundary

   n bdry pstop(rLo) peffective(rLo) pstop(rHi) peffective(rHi)
   5   -1      0.000           1.000      0.000           1.000
  10    1      0.244           0.756      0.025           0.975
  15    3      0.398           0.602      0.054           0.946
  ...
```
```

--------------------------------

### ph2single - Single-Stage Phase II Design

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates exact single-stage Phase II designs for clinical trials.

```APIDOC
## ph2single

### Description
Calculates exact single-stage Phase II designs when two-stage designs are not practical.

### Parameters
#### Arguments
- **pu** (numeric) - Required - Unacceptable response rate
- **pa** (numeric) - Required - Desirable response rate
- **ep1** (numeric) - Required - Type I error rate
- **ep2** (numeric) - Required - Type II error rate

### Request Example
ph2single(pu = 0.4, pa = 0.6, ep1 = 0.1, ep2 = 0.1)

### Response
#### Success Response (200)
- **n** (integer) - Sample size
- **r** (integer) - Response threshold
- **Type I** (numeric) - Actual Type I error
- **Type II** (numeric) - Actual Type II error
```

--------------------------------

### aucVardiTest - Compare Growth Curves

Source: https://context7.com/veseshan/clinfun/llms.txt

Performs two-sample tests for comparing growth curves using the area under the curve (AUC) method.

```APIDOC
## aucVardiTest

### Description
Compares growth curves between two groups using area under the curve (AUC) statistics.

### Method
`aucVardiTest`

### Endpoint
None (function call)

### Parameters
#### Path Parameters
None

#### Query Parameters
- **y** (numeric vector) - The response variable (e.g., tumor size).
- **time** (numeric vector) - The time points at which measurements were taken.
- **group** (integer vector) - The group indicator for each observation (e.g., 1 for control, 2 for treatment).

### Request Example
```r
library(clinfun)

# Example data
set.seed(111)
time <- rep(c(0, 7, 14, 21, 28), each = 10)
control <- 100 * exp(0.05 * time) + rnorm(50, sd = 10)
treatment <- 100 * exp(0.03 * time) + rnorm(50, sd = 10)

tumor_size <- c(control, treatment)
group <- rep(1:2, each = 50)
time_all <- rep(time, 2)

# Test for difference in growth curves
result <- aucVardiTest(
  y = tumor_size,
  time = time_all,
  group = group
)

print(result)
```

### Response
#### Success Response (200)
- **AUC Group 1** (numeric) - The calculated area under the curve for group 1.
- **AUC Group 2** (numeric) - The calculated area under the curve for group 2.
- **Z** (numeric) - The Z-statistic for the test.
- **p-value** (numeric) - The p-value associated with the Z-statistic.

#### Response Example
```
Two-sample test for growth curves

AUC Group 1: 3245.6
AUC Group 2: 2876.3

Z = 2.34, p-value = 0.019
```
```

--------------------------------

### Minimum Detectable Response Rate Difference (mdrr)

Source: https://context7.com/veseshan/clinfun/llms.txt

Computes minimum detectable response rates for marker-stratified designs.

```APIDOC
## mdrr - Minimum Detectable Response Rate Difference

### Description
Computes minimum detectable response rates for marker-stratified designs.

### Parameters
#### Request Body
- **n** (numeric) - Required - Total sample size.
- **cprob** (numeric) - Required - Proportion of marker-positive subjects.
- **presp** (numeric) - Required - Overall response probability.
- **alpha** (numeric) - Optional - Significance level (default is 0.05).
- **power** (numeric) - Optional - Statistical power (default is 0.8).

### Response
#### Success Response (200)
- **result** (list) - A list containing the minimum detectable response rates for marker-positive and marker-negative subjects.

### Request Example
```r
library(clinfun)
result <- mdrr(
  n = 200,
  cprob = 0.3,
  presp = 0.4,
  alpha = 0.05,
  power = 0.8
)
print(result)
```

### Response Example
```
Minimum detectable response rates
P(response | marker+): 0.589
P(response | marker-): 0.319
```
```

--------------------------------

### Design Group Sequential Trial for Binary Outcomes

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculate sample sizes and boundaries for group sequential designs with binary endpoints.

```r
library(clinfun)

# Design a 4-look group sequential trial for binary outcomes
```

--------------------------------

### twostage.inference - Statistical Inference After Two-Stage Trial

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates p-values, UMVUE estimates, and confidence intervals for completed two-stage trials.

```APIDOC
## twostage.inference

### Description
Calculates p-values, UMVUE estimates, and confidence intervals for completed two-stage trials.

### Parameters
#### Arguments
- **x** (integer) - Required - Observed responses
- **r1** (integer) - Required - Stage 1 threshold
- **n1** (integer) - Required - Stage 1 sample size
- **n** (integer) - Required - Total sample size
- **pu** (numeric) - Required - Null hypothesis response rate
- **alpha** (numeric) - Required - Significance level

### Response
#### Success Response (200)
- **UMVUE** (numeric) - Uniformly Minimum Variance Unbiased Estimator
- **p-value** (numeric) - Calculated p-value
- **CI** (numeric) - Confidence interval
```

--------------------------------

### toxbdry - Toxicity Monitoring Stopping Rule

Source: https://context7.com/veseshan/clinfun/llms.txt

Computes continuous stopping boundaries for toxicity monitoring during clinical trials.

```APIDOC
## toxbdry

### Description
Computes continuous stopping boundaries for toxicity monitoring during clinical trials.

### Method
`toxbdry(pLo, pHi, n, cP0, cP1)`

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
* **pLo** (numeric) - Acceptable toxicity rate under the null hypothesis.
* **pHi** (numeric) - Unacceptable toxicity rate under the alternative hypothesis.
* **n** (numeric vector) - Sample sizes at which to monitor toxicity.
* **cP0** (numeric) - Type I error rate (false positive rate) for stopping early for toxicity.
* **cP1** (numeric) - Power to detect high toxicity (1 - Type II error rate).

### Request Example
```r
library(clinfun)
tox_rule <- toxbdry(
  pLo = 0.2,
  pHi = 0.35,
  n = c(20, 40, 60, 75),
  cP0 = 0.1,
  cP1 = 0.9
)
print(tox_rule)

# Continuous monitoring example (every patient from 2 to 30)
continuous_rule <- toxbdry(pLo = 0.1, pHi = 0.3, n = 2:30)
print(continuous_rule)
```

### Response
#### Success Response (200)
Returns a matrix with columns for sample size (n), lower boundary, and upper boundary for toxicity.

#### Response Example
```
Toxicity boundary with cP(0.2) <= 0.1 and cP(0.35) >= 0.9

     n looks lo.bdry hi.bdry
[1,]      20       5       6
[2,]      40      11      12
[3,]      60      17      18
[4,]      75      22      23

Stop trial if number of toxicities exceeds boundary
```
```

--------------------------------

### gsdesign.binomial - Group Sequential Design for Binomial Outcomes

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates sample sizes and boundaries for group sequential trials with binomial outcomes (e.g., response rates).

```APIDOC
## gsdesign.binomial

### Description
Calculates sample sizes and boundaries for group sequential trials with binomial outcomes.

### Method
`gsdesign.binomial(ifrac, pC, pE, sig.level, power, delta.eb, delta.fb, alternative)`

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
* **ifrac** (numeric vector) - Information fractions at which to perform interim analyses.
* **pC** (numeric) - Response rate in the control arm.
* **pE** (numeric) - Response rate in the experimental arm.
* **sig.level** (numeric) - Significance level (Type I error rate).
* **power** (numeric) - Desired power of the trial.
* **delta.eb** (numeric) - Efficacy boundary parameter (e.g., O'Brien-Fleming).
* **delta.fb** (numeric) - Futility boundary parameter (e.g., O'Brien-Fleming).
* **alternative** (character) - Type of alternative hypothesis ('two.sided', 'less', 'greater').

### Request Example
```r
library(clinfun)
design <- gsdesign.binomial(
  ifrac = c(0.25, 0.5, 0.75, 1.0),
  pC = 0.2,
  pE = 0.35,
  sig.level = 0.05,
  power = 0.8,
  delta.eb = 0.5,
  delta.fb = 0.5,
  alternative = "two.sided"
)
print(design)
```

### Response
#### Success Response (200)
Returns a design object containing information fractions, efficacy boundaries, futility boundaries, and sample size per arm.

#### Response Example
```
Group sequential  design

Information fraction: 0.25 0.5 0.75 1
Efficacy boundary: 4.049 2.863 2.338 2.025
Futility boundary: 0.000 0.000 0.707 2.025
Sample size per arm: 112
```
```

--------------------------------

### Calculate Operating Characteristics of Two-Stage Boundary

Source: https://context7.com/veseshan/clinfun/llms.txt

Determine the operating characteristics for a specific two-stage design boundary.

```r
library(clinfun)

# Calculate operating characteristics for a specific design
oc <- oc.twostage.bdry(
  pu = 0.2,   # unacceptable response rate
  pa = 0.4,   # desirable response rate
  r1 = 3,     # stage 1 response threshold
  n1 = 19,    # stage 1 sample size
  r = 15,     # overall response threshold
  n = 54      # total sample size
)

print(oc)
#>        Type I   Type II   PET(p0)   EN(p0)
#>       0.04895   0.09982   0.58969   33.63
```

--------------------------------

### Perform Permutation Log-Rank Test (permlogrank)

Source: https://context7.com/veseshan/clinfun/llms.txt

The permlogrank function performs a log-rank test for survival comparisons, utilizing permutation reference distributions suitable for small samples. It requires survival objects and the number of permutations.

```r
library(clinfun)
library(survival)

# Small sample survival comparison
set.seed(456)
time <- c(rexp(15, 0.1), rexp(15, 0.2))    # survival times
status <- sample(0:1, 30, replace = TRUE)   # censoring status
group <- rep(1:2, each = 15)                # treatment groups

# Permutation log-rank test
result <- permlogrank(
  Surv(time, status) ~ group,
  nperm = 5000
)

print(result)
#> Permutation log-rank test
#>
#> Chi-square = 2.45, p-value = 0.124
#> (based on 5000 permutations)
```

--------------------------------

### Plot Survival Quantiles with Cox Model

Source: https://context7.com/veseshan/clinfun/llms.txt

Plots survival time quantile curves as a function of covariates using a fitted Cox proportional hazards model. Requires the survival and clinfun packages.

```r
library(clinfun)
library(survival)

# Fit Cox model
data(pbc, package = "survival")
pbc_subset <- subset(pbc, trt > 0 & copper > 0)

cox_fit <- coxph(
  Surv(time, status == 2) ~ trt + log(copper),
  data = pbc_subset
)

# Plot survival time vs covariate
plot(log(pbc_subset$copper), pbc_subset$time,
     pch = c("o", "x")[1 + (pbc_subset$status == 2)],
     xlab = "log(Copper)", ylab = "Survival Time (days)",
     main = "Median Survival by Copper Level")

# Add median survival quantile curve
coxphQuantile(cox_fit, xrange = c(2.5, 6), p = 0.5, whichx = 2, otherx = 1)

# Add 75th percentile curve in red
coxphQuantile(cox_fit, xrange = c(2.5, 6), p = 0.75, whichx = 2, otherx = 1, col = 2)
```

--------------------------------

### Compare ROC Curves (roc.area.test)

Source: https://context7.com/veseshan/clinfun/llms.txt

Tests for equality of areas under multiple ROC curves using DeLong's method.

```APIDOC
## roc.area.test - Compare ROC Curves

### Description
Tests for equality of areas under multiple ROC curves using DeLong's method.

### Parameters
#### Request Body
- **markers** (matrix or data frame) - Required - A matrix or data frame where columns represent different markers and rows correspond to subjects.
- **status** (vector) - Required - Vector of true status (0 for negative, 1 for positive).

### Response
#### Success Response (200)
- **result** (list) - A list containing the areas under the curve (AUCs) for each marker, their standard errors, and the results of the statistical test for equality of AUCs (Chi-square statistic, degrees of freedom, and p-value).

### Request Example
```r
library(clinfun)
set.seed(101)
status <- rep(0:1, each = 75)
marker_x <- rnorm(150) + 0.8 * status
marker_y <- rnorm(150) + 1.2 * status
marker_z <- rnorm(150) + 1.0 * status

result <- roc.area.test(
  markers = cbind(marker_x, marker_y, marker_z),
  status = status
)
print(result)
```

### Response Example
```
Areas under the curve:
        marker_x  marker_y  marker_z
Area      0.7142    0.8156    0.7623
SE        0.0423    0.0356    0.0391

Test for equality of AUCs:
Chi-square = 4.82, df = 2, p-value = 0.0898
```
```

--------------------------------

### fe.ssize - Fisher's Exact Sample Size

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates sample sizes based on Fisher's exact test for comparing two proportions.

```APIDOC
## fe.ssize

### Description
Calculates sample sizes based on Fisher's exact test for comparing two proportions.

### Method
`fe.ssize(p1, p2, alpha, power, r)`

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
* **p1** (numeric) - Proportion in group 1.
* **p2** (numeric) - Proportion in group 2.
* **alpha** (numeric) - Significance level (Type I error rate).
* **power** (numeric) - Desired power of the test.
* **r** (numeric) - Ratio of sample sizes between group 2 and group 1 (e.g., 1 for 1:1 randomization, 2 for 2:1).

### Request Example
```r
library(clinfun)
result <- fe.ssize(
  p1 = 0.2,
  p2 = 0.3,
  alpha = 0.05,
  power = 0.8,
  r = 1
)
print(result)

# With unequal allocation (2:1 ratio)
fe.ssize(p1 = 0.2, p2 = 0.35, alpha = 0.05, power = 0.9, r = 2)
```

### Response
#### Success Response (200)
Returns a data frame with calculated sample sizes per group and the corresponding exact power.

#### Response Example
```
             Group 1 Group 2 Exact Power
CPS              294     294   0.8080441
Fisher Exact     291     291   0.8006813
```
```

--------------------------------

### Compute Minimum Detectable Response Rate Difference (mdrr)

Source: https://context7.com/veseshan/clinfun/llms.txt

mdrr computes minimum detectable response rates for marker-stratified designs. It requires the total sample size, proportion of marker-positive individuals, overall response probability, alpha, and power.

```r
library(clinfun)

# What response rates can be detected by marker status?
# Given: n=200, 30% marker positive, 40% overall response
result <- mdrr(
  n = 200,           # total sample size
  cprob = 0.3,       # proportion marker positive
  presp = 0.4,       # overall response probability
  alpha = 0.05,
  power = 0.8
)

print(result)
#> Minimum detectable response rates
#> P(response | marker+): 0.589
#> P(response | marker-): 0.319
```

--------------------------------

### Permutation Log-Rank Test (permlogrank)

Source: https://context7.com/veseshan/clinfun/llms.txt

Performs log-rank test for survival comparisons using permutation reference distributions for small samples.

```APIDOC
## permlogrank - Permutation Log-Rank Test

### Description
Performs log-rank test for survival comparisons using permutation reference distributions for small samples.

### Parameters
#### Request Body
- **formula** (formula) - Required - A formula object, such as `Surv(time, status) ~ group`.
- **data** (data.frame) - Optional - A data frame containing the variables specified in the formula.
- **nperm** (integer) - Required - Number of permutations to perform.

### Response
#### Success Response (200)
- **result** (htest object) - An object of class 'htest' containing the test statistic, p-value, and other relevant information.

### Request Example
```r
library(clinfun)
library(survival)
set.seed(456)
time <- c(rexp(15, 0.1), rexp(15, 0.2))
status <- sample(0:1, 30, replace = TRUE)
group <- rep(1:2, each = 15)

result <- permlogrank(
  Surv(time, status) ~ group,
  nperm = 5000
)
print(result)
```

### Response Example
```
Permutation log-rank test

Chi-square = 2.45, p-value = 0.124
(based on 5000 permutations)
```
```

--------------------------------

### Design Group Sequential Trial for Binomial Outcomes

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates sample sizes for trials with binomial outcomes using gsdesign.binomial.

```r
design <- gsdesign.binomial(
  ifrac = c(0.25, 0.5, 0.75, 1.0),  # information fractions
  pC = 0.2,                         # control arm response rate
  pE = 0.35,                        # experimental arm response rate
  sig.level = 0.05,
  power = 0.8,
  delta.eb = 0.5,                   # O'Brien-Fleming efficacy boundary
  delta.fb = 0.5,                   # O'Brien-Fleming futility boundary
  alternative = "two.sided"
)

print(design)
```

--------------------------------

### Compare Growth Curves Using AUC

Source: https://context7.com/veseshan/clinfun/llms.txt

Performs a two-sample test for comparing growth curves based on the area under the curve (AUC). Requires the clinfun package and simulated data.

```r
library(clinfun)

# Compare tumor growth curves between treatment groups
set.seed(111)
# Time points
time <- rep(c(0, 7, 14, 21, 28), each = 10)

# Control group: faster growth
control <- 100 * exp(0.05 * time) + rnorm(50, sd = 10)

# Treatment group: slower growth
treatment <- 100 * exp(0.03 * time) + rnorm(50, sd = 10)

# Combine data
tumor_size <- c(control, treatment)
group <- rep(1:2, each = 50)
time_all <- rep(time, 2)

# Test for difference in growth curves
result <- aucVardiTest(
  y = tumor_size,
  time = time_all,
  group = group
)

print(result)
```

--------------------------------

### Minimum Detectable Odds Ratio (fe.mdor)

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates the minimum detectable odds ratio given sample sizes, proportion in control group, alpha, and power.

```APIDOC
## fe.mdor - Minimum Detectable Odds Ratio

### Description
Calculates the minimum detectable odds ratio given sample sizes.

### Parameters
#### Request Body
- **ncase** (numeric) - Required - Number of cases.
- **ncontrol** (numeric) - Required - Number of controls.
- **pcontrol** (numeric) - Required - Proportion in control group.
- **alpha** (numeric) - Optional - Significance level (default is 0.05).
- **power** (numeric) - Optional - Statistical power (default is 0.8).

### Response
#### Success Response (200)
- **result** (list) - A list containing the minimum detectable odds ratio calculated using different methods (Schlesselman, CPS, Fisher Exact) along with their exact power.

### Request Example
```r
library(clinfun)
result <- fe.mdor(
  ncase = 100,
  ncontrol = 200,
  pcontrol = 0.3,
  alpha = 0.05,
  power = 0.8
)
print(result)
```

### Response Example
```
             Min OR Exact Power
Schlesselman  2.163   0.8000000
CPS           2.098   0.8009932
Fisher Exact  2.080   0.8014276
```
```

--------------------------------

### coxphQuantile - Survival Quantiles by Covariate

Source: https://context7.com/veseshan/clinfun/llms.txt

Draws survival time quantile curves as a function of covariates using a fitted Cox proportional hazards model.

```APIDOC
## coxphQuantile

### Description
Draws survival time quantile curves as a function of covariates.

### Method
`coxphQuantile`

### Parameters
#### Path Parameters
None

#### Query Parameters
- **cox_fit** (Coxph object) - The fitted Cox proportional hazards model object from `survival::coxph`.
- **xrange** (numeric vector) - The range of covariate values for plotting.
- **p** (numeric) - The quantile to plot (e.g., 0.5 for median).
- **whichx** (numeric) - The index of the covariate in the `cox_fit` model to vary.
- **otherx** (numeric vector, optional) - Fixed values for other covariates in the model.
- **col** (character or numeric, optional) - Color for the plotted curve.

### Request Example
```r
library(clinfun)
library(survival)

# Fit Cox model
data(pbc, package = "survival")
pbc_subset <- subset(pbc, trt > 0 & copper > 0)

cox_fit <- coxph(
  Surv(time, status == 2) ~ trt + log(copper),
  data = pbc_subset
)

# Add median survival quantile curve
coxphQuantile(cox_fit, xrange = c(2.5, 6), p = 0.5, whichx = 2, otherx = 1)

# Add 75th percentile curve in red
coxphQuantile(cox_fit, xrange = c(2.5, 6), p = 0.75, whichx = 2, otherx = 1, col = 2)
```

### Response
This function generates a plot and does not return a value directly. The plot displays survival quantile curves.
```

--------------------------------

### Calculate Fisher's Exact Sample Size

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates sample sizes based on Fisher's exact test for comparing two proportions using fe.ssize.

```r
library(clinfun)

# Calculate sample size for comparing response rates:
# - Standard treatment: 20% response
# - Experimental: 30% response
# - 80% power, 5% alpha
result <- fe.ssize(
  p1 = 0.2,      # response rate group 1
  p2 = 0.3,      # response rate group 2
  alpha = 0.05,
  power = 0.8,
  r = 1          # 1:1 randomization
)

print(result)

# With unequal allocation (2:1 ratio)
fe.ssize(p1 = 0.2, p2 = 0.35, alpha = 0.05, power = 0.9, r = 2)
```

--------------------------------

### Design Group Sequential Trial for Normal Outcomes

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates sample sizes for trials comparing means with normally distributed outcomes using gsdesign.normal.

```r
library(clinfun)

# Design a 3-look group sequential trial for continuous outcomes
design <- gsdesign.normal(
  ifrac = c(1/3, 2/3, 1),      # look at 1/3, 2/3, and full enrollment
  delta = 0.5,                  # true difference in means
  sd = 1,                       # standard deviation
  sig.level = 0.05,
  power = 0.9,
  delta.eb = 0.5,               # O'Brien-Fleming boundary
  alternative = "two.sided"
)

print(design)
```

--------------------------------

### Calculate Minimum Detectable Odds Ratio (fe.mdor)

Source: https://context7.com/veseshan/clinfun/llms.txt

Use fe.mdor to determine the minimum detectable odds ratio given sample sizes for cases and controls, along with the proportion in the control group, alpha, and power.

```r
library(clinfun)

# Given sample sizes, what odds ratio can be detected?
result <- fe.mdor(
  ncase = 100,       # cases
  ncontrol = 200,    # controls
  pcontrol = 0.3,    # proportion in control group
  alpha = 0.05,
  power = 0.8
)

print(result)
#>              Min OR Exact Power
#> Schlesselman  2.163   0.8000000
#> CPS           2.098   0.8009932
#> Fisher Exact  2.080   0.8014276
```

--------------------------------

### Perform Jonckheere-Terpstra Test (jonckheere.test)

Source: https://context7.com/veseshan/clinfun/llms.txt

Use jonckheere.test for exact or permutation-based Jonckheere-Terpstra tests to detect ordered differences among groups. Specify the alternative hypothesis as 'increasing' or 'decreasing'. For tied data, permutation tests are recommended.

```r
library(clinfun)

# Test for increasing trend across dose groups
set.seed(123)
dose_group <- rep(1:4, each = 15)          # 4 dose levels
response <- rnorm(60) + 0.3 * dose_group   # increasing trend

result <- jonckheere.test(
  x = response,
  g = dose_group,
  alternative = "increasing"
)

print(result)
#> Jonckheere-Terpstra test
#>
#> data:  response by dose_group
#> JT = 1078, p-value = 0.0009213
#> alternative hypothesis: increasing

# With tied data, use permutation test
response[1:3] <- mean(response[1:3])  # create ties
result_perm <- jonckheere.test(
  x = response,
  g = dose_group,
  alternative = "increasing",
  nperm = 5000
)
print(result_perm)
```

--------------------------------

### Concordance Probability Estimate (coxphCPE)

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates the Gonen-Heller concordance probability estimate for Cox regression models.

```APIDOC
## coxphCPE - Concordance Probability Estimate

### Description
Calculates the Gonen-Heller concordance probability estimate for Cox regression models.

### Parameters
#### Request Body
- **cox_fit** (coxph object) - Required - A fitted Cox proportional hazards model object (from `survival::coxph`).
- **out.ties** (logical) - Optional - If TRUE, handles ties in risk scores by averaging. Defaults to FALSE.

### Response
#### Success Response (200)
- **cpe** (numeric vector) - A vector containing the Concordance Probability Estimate (CPE), its smoothed value, and its standard error.

### Request Example
```r
library(clinfun)
library(survival)

data(pbc, package = "survival")
pbc_subset <- subset(pbc, trt > 0 & copper > 0)

cox_fit <- coxph(
  Surv(time, status == 2) ~ age + log(copper) + log(bili),
  data = pbc_subset
)

cpe <- coxphCPE(cox_fit)
print(cpe)

# For discrete risk groups, use out.ties = TRUE
cpe_discrete <- coxphCPE(cox_fit, out.ties = TRUE)
```

### Response Example
```
     CPE smooth.CPE    se.CPE
  0.7234     0.7189    0.0298
```
```

--------------------------------

### Empirical ROC Curve (roc.curve)

Source: https://context7.com/veseshan/clinfun/llms.txt

Computes and plots empirical ROC curves for diagnostic markers.

```APIDOC
## roc.curve - Empirical ROC Curve

### Description
Computes and plots empirical ROC curves for diagnostic markers.

### Parameters
#### Request Body
- **marker** (numeric vector) - Required - Vector of marker values.
- **status** (vector) - Required - Vector of true status (0 for negative, 1 for positive).
- **nboot** (integer) - Optional - Number of bootstrap samples for confidence intervals. Defaults to 0 (no bootstrapping).
- **PRC** (logical) - Optional - If TRUE, computes and plots a Precision-Recall Curve instead of an ROC curve. Defaults to FALSE.

### Response
#### Success Response (200)
- **result** (list) - A list containing AUC, standard error, and optionally confidence intervals. If PRC is TRUE, it returns precision-recall curve data.

### Request Example
```r
library(clinfun)
set.seed(789)
status <- rep(0:1, each = 50)
marker1 <- rnorm(100) + status
marker2 <- rnorm(100) + 1.5 * status

roc1 <- roc.curve(marker = marker1, status = status)
roc2 <- roc.curve(marker = marker2, status = status)

print(roc1)
print(roc2)

# Plot ROC curves
plot(roc1, main = "ROC Curve Comparison")
lines(roc2, col = "red")
legend("bottomright", c("Marker 1", "Marker 2"), col = c("black", "red"), lty = 1)

# Plot Precision-Recall curve
plot(roc1, PRC = TRUE, main = "Precision-Recall Curve")
```

### Response Example
```
Area under the curve: 0.7368
Standard error: 0.0489

Area under the curve: 0.8624
Standard error: 0.0372
```
```

--------------------------------

### Compare ROC Curve Areas (roc.area.test)

Source: https://context7.com/veseshan/clinfun/llms.txt

Use roc.area.test to test for equality of areas under multiple ROC curves using DeLong's method. It requires a matrix of marker values and the corresponding disease status.

```r
library(clinfun)

# Compare multiple diagnostic markers
set.seed(101)
status <- rep(0:1, each = 75)
marker_x <- rnorm(150) + 0.8 * status
marker_y <- rnorm(150) + 1.2 * status
marker_z <- rnorm(150) + 1.0 * status

# Test if AUCs are significantly different
result <- roc.area.test(
  markers = cbind(marker_x, marker_y, marker_z),
  status = status
)

print(result)
#> Areas under the curve:
#>         marker_x  marker_y  marker_z
#> Area      0.7142    0.8156    0.7623
#> SE        0.0423    0.0356    0.0391
#>
#> Test for equality of AUCs:
#> Chi-square = 4.82, df = 2, p-value = 0.0898
```

--------------------------------

### gsdesign.normal - Group Sequential Design for Normal Outcomes

Source: https://context7.com/veseshan/clinfun/llms.txt

Calculates sample sizes for group sequential trials comparing means with normally distributed outcomes.

```APIDOC
## gsdesign.normal

### Description
Calculates sample sizes for trials comparing means with normally distributed outcomes using a group sequential design.

### Method
`gsdesign.normal(ifrac, delta, sd, sig.level, power, delta.eb, alternative)`

### Parameters
#### Path Parameters
None

#### Query Parameters
None

#### Request Body
* **ifrac** (numeric vector) - Information fractions at which to perform interim analyses.
* **delta** (numeric) - True difference in means between the two groups.
* **sd** (numeric) - Standard deviation of the outcome.
* **sig.level** (numeric) - Significance level (Type I error rate).
* **power** (numeric) - Desired power of the trial.
* **delta.eb** (numeric) - Efficacy boundary parameter (e.g., O'Brien-Fleming).
* **alternative** (character) - Type of alternative hypothesis ('two.sided', 'less', 'greater').

### Request Example
```r
library(clinfun)
design <- gsdesign.normal(
  ifrac = c(1/3, 2/3, 1),
  delta = 0.5,
  sd = 1,
  sig.level = 0.05,
  power = 0.9,
  delta.eb = 0.5,
  alternative = "two.sided"
)
print(design)
```

### Response
#### Success Response (200)
Returns a design object containing information fractions, efficacy boundaries, and sample size per arm.

#### Response Example
```
Group sequential design
Information fraction: 0.333 0.667 1.000
Efficacy boundary: 3.471 2.454 2.004
Sample size per arm: 86
```
```