diff --git a/README.md b/README.md index 9458830..28a1f0e 100644 --- a/README.md +++ b/README.md @@ -1,8 +1,8 @@ # Wonderful Wednesdays -**April 10th, 2024**. +**May 08th, 2024**. -The challenge for the next webinar is about study flow charts. Details can be found [here](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-04-10). +The challenge for the next webinar is about platform trial design. Details can be found [here](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-05-08). For more on PSI and wonderful wednesdays see https://www.psiweb.org/sigs-special-interest-groups/visualisation/welcome-to-wonderful-wednesdays @@ -20,9 +20,9 @@ You will be able to submit your improvements for feedback via a google form on t We will make the submissions available to the community together with highlights of the strength and limitations through our [blog](https://vis-sig.github.io/blog/). Over time, this will lead to a gallery of visualizations for others to learn from. -## April data set +## May data set -**Upcoming May 8th, 2024 webinar**. The current data example focuses on [study flow charts](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-04-10). +**Upcoming June 12th, 2024 webinar**. The current data example focuses on [platform trial designs](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-05-08). Submit your visualisations [here](https://docs.google.com/forms/d/e/1FAIpQLSdNAaiuUAD89LAdQm5KNnLWs-MjqA4pzX2VHAwN7iqwoKpi-Q/viewform) or send them to ! @@ -122,6 +122,8 @@ Submit your visualisations [here](https://docs.google.com/forms/d/e/1FAIpQLSdNAa * [Power and sample size calculations](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-03-13). Presented on **March 13th, 2024**. +* [Study flow charts](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-04-10). Presented on **April 10th, 2024**. + diff --git a/data/2024/2024-05-08/Readme.md b/data/2024/2024-05-08/Readme.md new file mode 100644 index 0000000..26d1d6e --- /dev/null +++ b/data/2024/2024-05-08/Readme.md @@ -0,0 +1,40 @@ +# Background + +- It is known that sometimes at the design stage of a clinical trial, simulations are conducted to better understand the operating characteristics of the design. +- In a [recently published paper](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9304586/), a simulation study was conducted for a platform trial design in the context of NASH (Non-Alcoholic Steatohepatitis). Different assumptions regarding e.g. treatment effect and correlation between the surrogate interim endpoint and the final endpoint, as well as design choices such as sample size and maximum number of cohorts allowed in the platform were investigated. + + +# Data + +- The simulated dataset can be found on [Github](https://github.com/el-meyer/airship/blob/master/data/ExampleDataNASH.csv). +- Simulation Input Columns and their meaning: + - ShortTermEndpointQuality = Correlation between surrogate interim and final endpoint + - Maximumnumberofcohorts = Maximum number of cohorts allowed to enter the platform + - TypeofDataSharing = How is data shared across cohorts + - CohortInclusionRate = Rate at which new cohorts enter the platform (higher levels indicate faster entry) + - FinalCohortSampleSize = Final Sample Size per Cohorts + - InterimFutilityStopping = Binding early futility stopping yes/no + - TreatmentEfficacySetting = Treatment effect scenario (can be found in paper). Often the treatment effect assignment is random following a distribution. +- Simulation Output Columns and their meaning: + - Avg_Pat = Average number of participants enrolled in the trial + - Avg_Pat_First_Suc = Average number of participants enrolled in the trial until the first treatment is declared successful + - Avg_Perc_Pat_Sup_Plac = Average percentage of patients who receive a treatment that is better than control + - Avg_Cohorts = Average number of cohorts enrolled in the trial + - FDR = False Discovery Rate + - PTP = Per-Cohort-Power; NA in pure null scenarios + - PTT1ER = Per-Cohort-Type 1 Error; NA in pure alternative scenarios + - FWER = Family-Wise-Error-Rate + - Disj_Power = Disjunctive Power + - FWER_BA = Family Wise Error Rate where only simulations are used in the denominator that contained a treatment that is not superior to control + - Disj_Power_BA = Disjunctive Power where only simulations are used in the denominator that contained a treatment that is superior to control + + +# Challenge + +- How can we best show the effect of sample size on the power/type 1 error? +- Does sample size affect PTP and Disj_Power in the same way? How about in interaction with the type of data sharing? +- What is the relationship between number of cohorts, cohort inclusion rate and power? +- How can we investigate the effectiveness of early stopping for futility? +- What is the relationship between the quality of the surrogate endpoint and the average number of patients enrolled? Is there an interaction with futility stopping and treatment effect scenario? +- How do FWER and FWER_BA differ from each other with respect to treatment effect scenario and maximum number of cohorts? +- Can you find other interesting patterns in the data? diff --git a/data/Readme.md b/data/Readme.md index 3a5e5df..3b8f8b1 100644 --- a/data/Readme.md +++ b/data/Readme.md @@ -111,6 +111,8 @@ Table of contents for the 2023 webinar series data sets: - [Power and sample size calculations](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-03-13). Presented on **March 13th, 2024**. - [Study flow charts](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-04-10). Presented on **April 10th, 2024**. + + - [Platform trial designs](https://github.com/VIS-SIG/Wonderful-Wednesdays/tree/master/data/2024/2024-05-08). **NOTE** to download a single data set as a csv file, click on the raw button and save the file. The [following link describes the process in