interval_auc_2curve: Single Predictor, Difference Between Two ROC Curve Confidence...

Description Usage Arguments Details Value Author(s) References

View source: R/interval_auc_2curve.R

Description

Calculates the sample size, AUC's, margin of error, or significance for the confidence interval of the difference of two Areas Under the Curve (AUC). Assumes that the AUCs are based on a single numeric predictor and that the AUCs are normally distributed.

Usage

1
2
interval_auc_2curve(auc1 = NULL, auc2 = NULL, n = NULL, E = NULL,
  alpha = 0.05, r, weights = list(c(1, 1)), optim.max = 1e+09)

Arguments

auc1

Observed Area Under the Curve 1 (AUC)

auc2

Observed Area Under the Curve 2 AUC)

n

Total sample size

E

Desired margin of error

alpha

Significance Level

r

The correlation introduced between the two AUCs by studying the same sample of patients.

weights

a list of length-2 vectors giving the weights for the diseased and non-diseased groups. This is used to calculate k, the proportion of subjects belonging to the first group.

optim.max

Maximum value for optimze when solving for n. The default setting is 1,000,000,000, which is probably higher than anyone will ever need. I may change this in the future for the sake of improving speed.

Details

The ROC curves for this comparison should be based on the same sample. That is, it should be two tests performed on the same sample. The comparison assumes some dependence between the curves, with this dependence being defined by r.

Exactly one of the parameters n, E, auc1, auc2 and alpha must be passed as NULL, and that parameter will be calculated from the others. Notice that alpha has a non-NULL default, so NULL must be explicitly passed if you want it computed.

Value

Returns a data frame of the parameters passed to the function (expanded using expand.grid) and the corresponding sample size estimates.

  1. E: Margin of Error

  2. auc1: Area Under the Curve 1 (AUC)

  3. auc2: Area Under the Curve 2 (AUC)

  4. k: Proportion of total sample size belonging to group 1

  5. r: The correlation introduced between the two AUCs by studying the same sample of patients.

  6. n1_est: Exact sample size for group 1

  7. n2_est: Exact sample size for group 2

  8. n_est: Exact total sample size

  9. n1: Estimated sample size for group 1 (next largest integer)

  10. n2: Estimated sample size for group 2 (next largest integer)

  11. n: Estimated total sample size (sum of n1 and n2)

Author(s)

Benjamin Nutter

References

James A Hanley and Barbara J McNeil, "A Method of Comparing the Areas Under Receiver Operating Characteristic Curves Derived from the Same Cases," Radiology, Vol. 148, No. 3, Pages 839-843, September 1983.

Nancy A Obuchowski, "Sample size calculations in studies of test accuracy," Statistical Methods in Medical Research 1998; 7: 371-392

James A. Hanley and Barbara J. McNeil, "The Meaning and Use of the Area under a Receiver Operating Characteristic (ROC) Curve," Radiology. Vol 143. No 1, Pages 29-36, April 1982.


nutterb/junkyard documentation built on May 24, 2019, 10:51 a.m.