Synthetic biomarker study: the planted effects and their true sizes

The three effects planted in the synthetic study, each with its true size. This list is the generator's own parameter list: the script data-raw/synthetic-study.R read these values to make the data, so a test that asserts against them asserts against the truth and not a copy of it.

Usage

Synthetic_Truth

Format

A list of four:

Seed

integer The seed the study was generated with.

GroupDifference

A list: Biomarker; Visit, the visit tests use, and BaselineVisit; GroupCol, the column of Synthetic_Participants that holds the groups; Groups, in the order of the subtraction; Estimand, in words; and Value, the true difference.

Correlation

A list: Biomarkers, the pair; Visit, the visit tests use; Method, "pearson"; Estimand; and Value, the true coefficient.

Survival

A list: Biomarker; Visit; Cut, "median"; Groups, numerator first; Endpoint, the PARAMCD in Synthetic_Outcomes; Estimand; and Value, the true hazard ratio.

Source

Written by data-raw/synthetic-study.R.

Details

Each true size is an exact parameter of the generating model, stated in the terms the matching statistic estimates:

Effect	Where	Truth
Difference in mean change from Baseline, Treatment minus Placebo	IL-6, at each visit after Baseline	-1.5 pg/mL
Pearson correlation	TNF-alpha and IL-10, at each visit	0.6
Hazard ratio, high against low	Baseline CRP split at its median, event-free survival	2.5

Every other biomarker is null by construction. The study was generated once, with seed 364, and each truth lies inside the 95 percent interval base R computes from the data.

See also

Synthetic_Results, Synthetic_Participants, Synthetic_Outcomes