Custom software development
Mediana has developed custom software tools to support the implementation and simulation-based evaluation of innovative trial designs and analysis methods, e.g., multiplicity adjustment and subgroup analysis methods.
The tools rely on advanced simulation engines written in C++ that are 50-100 times faster than R or SAS code and can be further enhanced by using parallel computing solutions on computers or servers with multiple cores. A graphical user interface can be constructed using Microsoft Visual Studio to build Windows-based application or an extended Shiny package to create web-based applications. We have also built custom R packages for our clients. The applications or packages support useful features such as an automatic generation of comprehensive simulation reports (Microsoft Word documents) or slide sets (Microsoft PowerPoint presentations). Web applications are released on our cloud platform that takes advantage of the power of the Amazon Cloud.
Custom software tools have been developed to support the following types of adaptive designs:
- Bayesian adaptive designs in early-stage clinical trials: Adaptive designs based on the modified toxicity probability interval (mTPI) method and Bayesian logistic regression models (BLRM) in dose-escalation trials with a single agent and combination of two agents.
- Adaptive designs with sample size re-estimation: Adaptive designs for confirmatory Phase III trials that support an option to update the total number of patients or target number of events based on the interim analysis results and also support early stopping due to futility or superior efficacy.
- Adaptive designs with arm selection: Adaptive designs for confirmatory Phase III trials that support an option to select a fixed number of best trial arms, e.g., doses with the strongest treatment effect, based on the interim analysis results. The designs also support sample size re-estimation and early stopping due to futility or superior efficacy.
- Adaptive designs with population selection: Adaptive designs for confirmatory Phase III trials with several pre-defined patient populations that support population and hypothesis selection rule to identify the most appropriate patient population and analysis strategy based on the interim analysis results. The designs also support sample size re-estimation and early stopping due to futility or superior efficacy.
Custom software tools have been built to support multiplicity adjustment methods in the following settings:
- Traditional multiplicity problems: Multiplicity adjustment methods for settings with a single source of multiplicity, e.g., trials with multiple efficacy endpoints.
- Advanced multiplicity problems: Multiplicity adjustment methods for complex settings with several sources of multiplicity, e.g., trials with multiple endpoints evaluated at several doses of a novel treatment or multiple endpoints evaluated in several patient populations.
- Advanced multiplicity problems in group-sequential or adaptive trials: Multiplicity adjustment methods for complex settings with several sources of multiplicity, e.g., trials with multiple efficacy endpoints evaluated at several doses with multiple decision points (interim and final analyses).
Custom software tools have been developed to support subgroup analysis or subgroup exploration approaches in confirmatory and exploratory settings:
- Confirmatory subgroup analysis: Efficacy evaluations in multi-population settings, i.e., trials with a small number of pre-specified patient subpopulations in addition to the overall population.
- Exploratory subgroup analysis: Efficient subgroup searches in trials with a negative overall outcomes based on a set of candidate biomarkers.