FACTS 6.5.0 Release Notes
November 21, 2022
1 Introduction
FACTS 6.5.0 is now available for download via App Center. Please contact us regarding any questions.
FACTS users can now:
- Specify frequentist margins (“deltas”) in the calculation of p-value and predictive probability QOIs for FACTS Core and Staged designs (except Time-to-Event designs).
- Create designs with interims triggered based on predictor events for FACTS Core and Staged Time-to-Event designs with a Time-to-Event predictor.
- Create designs where the final event endpoint analysis can be performed without any imputation based on the predictor endpoint for FACTS Core and Staged Time-to-Event designs with a predictor endpoint.
- Observe significant improvements in the mixing of MCMC chains within the Bayesian Augmented Control and Active Comparator (BAC and BAAC respectively) hierarchical models for FACTS Core and Staged and Enrichment designs.
- Generate design reports for FACTS Core Multiple Endpoint designs, FACTS Enrichment designs (Continuous, Dichotomous and Time-to-Event) and FACTS N-CRM designs.
2 FACTS Core and Staged Improvements
- In FACTS Core and Staged designs (except Time-to-Event designs), FACTS users can now globally specify a frequentist super-superiority/non-inferiority margin on the Quantities of Interest tab under the Standard Evaluation Variables area, which will be applied to the calculation of all p-value QOIs and “Current Trial” Predictive Probability QOIs. Note that this globally defined margin does not apply to “Future Trial” Predictive Probability QOIs, which can have their own separate margin defined. In addition, users now have the option on the “Frequentist Analysis” tab to use the frequentist super-superiority/non-inferiority margin in the frequentist analysis.
- In FACTS Core and Staged Time-to-Event designs with a Time-to-Event predictor, users can now specify designs with interims triggered based on the number of predictor events that have been observed. In addition, and independently of how interims are triggered, users can now specify maximum event caps based on either Final events or Predictor events.
- In FACTS Core and Staged Time-to-Event designs with a predictor endpoint, users can now specify the final endpoint analysis to not depend on any imputation from the predictor endpoint. This can be achieved by selecting the “No imputation” option within the “Imputation on Predictor” panel on the Design > Predictor Model > Relationship to Endpoint tab.
- In FACTS Core and Staged designs, the mixing of MCMC chains within Bayesian Augmented Control and Active Comparator (BAC and BAAC respectively) hierarchical models has been improved significantly; in particular, when the prior on tau^2 in the hierarchical model is forced to be small.
- FACTS Core Multiple Endpoint now provides the ability to generate a design report once the design has been simulated. As a result, all FACTS Core design types can now generate design reports.
- FACTS Core and Staged designs now correctly display a trial as having stopped for futility if all arms have been dropped.
- FACTS Core and Staged designs now correctly prevent interims from being performed beyond full enrollment when the “Discontinue interim analysis beyond full enrolment” setting on the Interims tab is selected.
- FACTS Staged Time-to-Event designs now correctly handle interim timings in Stage 2 for the various data inclusion rules as specified on the Data Inclusion tab, and interim information based on “Just Stage 2 data” or “Stage 2 and included Stage 1 data”, as specified on the Stage 2 Interims tab.
- FACTS Staged Time-to-Event designs now correctly handles interim timings based on complete predictor data, when a predictor is included in the design and the “Primary endpoint is censoring for intermediate predictor” setting is selected.
- FACTS Core and Staged Time-to-Event designs now correctly handle predictor based imputation when using a dichotomous predictor endpoint.
- On the Analysis tab in FACTS Core and Staged Time-to-Event designs, current trial predictive probabilities that estimate an accrual rate no longer require input data to be sorted by accrual time.
- FACTS Core Multiple Endpoint and FACTS Staged Dichotomous designs will now correctly output p-value trend test QOIs as a single output column, rather than one output column per dose, in summary files.
- FACTS Staged Multiple Endpoint designs will now correctly display the endpoint number when outputting p-value trend test QOIs in summary files.
- FACTS (Staged) Multiple Endpoint designs will now correctly display the posterior probability QOI comparison options (“Rates” and “Log-odds”) for dichotomous endpoints. Changing the endpoint from being dichotomous to continuous will delete posterior probability QOIs using the “Log-odds” comparison.
- FACTS Core and Staged Multiple Endpoint analyses will now correctly handle endpoints with visits schedules. Namely, when an endpoint contains only one visit schedule or when an endpoint’s visit schedule involved non-consecutive visits.
- FACTS Core and Staged Multiple Endpoint analyses will now correctly handle endpoints whose visit schedule contains missing data.
3 FACTS Enrichment Design Improvements
- FACTS Enrichment designs (Continuous, Dichotomous and Time-to-Event) now provide the ability to generate design reports once the designs have been simulated. As a result, all FACTS Enrichment design types can now generate design reports.
- The mixing of MCMC chains within Bayesian Augmented Control (BAC) hierarchical model has been improved significantly; in particular, when the prior on tau^2 in the hierarchical model is forced to be small.
- The clustering in Enrichment designs has been improved for situations when the prior on tau^2 is chosen to be small (e.g. 0.01 with weight 1).
- The patients file output from simulations (patients.csv) now correctly populates the dropout state of patients, and can now be used as subject data input on the Analysis tab without requiring modification.
- MCMC Trace plots are now available for all Enrichment design types when viewing simulation results graphs and when performing analyses. To view these graphs, at least one MCMC file needs to be generated. This can be done by going to the Simulations tab > MCMC Settings.
- External data file validation has been improved.
4 FACTS Dose Escalation Improvements
- N-CRM now provides the ability to generate design reports once the design has been simulated.
- In N-CRM designs which include efficacy, the “Maximum cohorts used to determine MTD” setting on the Allocation Rule tab is now observed correctly.
- In N-CRM designs, when deriving toxicity/efficacy priors from quantiles at the lowest and highest doses, the “middle” probability option now refers to the median dose rather than the reference dose and is now optional. When pre-6.5 FACTS files are loaded in FACTS 6.5, the “middle” probability option will be ignored and replaced by 0.5.
- In N-CRM designs, the specification of at least two dose levels is now required when deriving toxicity/efficacy priors from specific quantiles. Previously, the specification of at least three dose levels was required.
- FACTS Command Line mode and FLFLL (FACTS Linux File Loader Lite) now correctly handle the processing of N-CRM designs whose prior was derived using the “Legacy Prior” option.
- In N-CRM designs using open enrollment without accrual pausing, FACTS will now output entries in the cohort file for subjects that have been lost following stopping rules being met.
- In N-CRM designs using open enrollment, FACTS will no longer prematurely apply stopping conditions when there are still subjects whose outcomes have been observed, but not yet processed.
- In N-CRM designs using open enrollment, FACTS will now report the selected MTD/MED/OSD and associated state to be the dose where the stopping conditions were first met, unless the MTD has subsequently decreased post stopping rules being met.
- In N-CRM designs using open enrollment, dose escalation rules when using fine-grained dosing will behave correctly.
- In N-CRM designs using open enrollment and two groups, stopping rules and dose escalations rules will now apply to the correct group.
5 General Improvements
- FACTS now targets .NET Framework 4.8, the latest major version of the .NET Framework.
- A new “Simulation Duration” table can be viewed when right-clicking on a simulation design scenario. The Simulation Duration table gives a granular view of simulation start and end times, as well as its total duration.
- Several major improvements to FLFLL (enterprise licensees only): in particular, the ability to process specific scenarios of a design, the ability to process all FACTS files contained within a specified directory, and the reporting of design scenario validation errors. See FLFLL documentation for details.
- In FACTS Command Line mode and FLFLL, a new flag is available to specify the number of MCMC samples to generate for imputation purposes.
- In FACTS Command Line mode, the ability to generate a design report has been added. This can be achieved by adding the -report flag and the -rpath flag, where the latter is used to specify the path to the R executable.
- FACTS now provides links to FACTS introductory videos hosted on YouTube via the Help menu.
- Simulation engine errors in FACTS are now displayed in the GUI more informatively.
- The remaining time left on a FACTS license is now displayed correctly on the FACTS splash screen and Help menu.
- FACTS can now output up to 99,999 patients/weeks/frequentist/MCMC files. Previously, this was capped at 9,999 files.
- All designs supporting design report generation can have their design report generated without having to perform an additional command execution step in RStudio, by selecting a valid R installation under Settings > Options > R Configuration.
- FACTS will now correctly handle the serialization/deserialization of text inputs involving the following characters: “>”, “<”, “&”, “ ’ ” and “\”.
- When viewing FACTS simulation results through the GUI, estimates of responses, effects and hazard ratios (for Time-to-Event designs) will be display more obviously in the results column headers.