Decision support tools for clinical trial design

摘要

Many published clinical trials are poorly designed, suggesting that the protocol was incomplete, disorganised, or contained errors. This fact motivated the development of a suite of decision support tools for the design of randomised controlled clinical trials. In this paper we describe these tools, discussing both underlying theoretical issues and usage of the tools. The core tool—Design-a-Trial (DaT)—critiques data entered so as to guide design of a scientifically and ethically sound trial. DaT outputs a text protocol describing the trial, and a corresponding symbolic representation. Linked to DaT is a tool for authoring plans that form part of the trial. A key feature of this tool is the provision of macros for describing commonly occurring plan constructs. We describe another linked tool which generates solutions to Prolog queries requesting advice on how a plan should be revised so as to comply with safety and efficacy requirements. The user is able to navigate a path through the solution search space by interacting with natural language representations of the Prolog sub-goals. This provides the flexibility to generate useful and informative partial solutions, symbolic and textual, for inclusion in the symbolic plan representation and protocol document, respectively.