Aims and objective
The Experimental Design Assistant (EDA) is a web application which helps researchers improve the design of animal experiments. Carefully designed experiments yield robust and reproducible data using the minimum number of animals consistent with the scientific objectives.
Through a graphical interface, the EDA allows experiments to be designed online, as a diagram, using a practical, intuitive and flexible tool supported by web-based resources.
Audience and benefits
The resource is aimed at scientists who use animals in their research. Benefits include advice and feedback on the experimental plans, along with a range of functionalities providing support with the randomisation and blinding of the experiment, as well as sample size calculation. It equips researchers with practical information and knowledge, allowing them to determine the most efficient design for their experiment and understand the implications of choosing a particular design.
The EDA is not designed to replace specialist statistical advice. Depending on the level of expertise on experimental design and statistics, the EDA can be used in different ways.
For researchers who have limited access to statistical support, the feedback and advice provided by the system will be particularly pertinent, as it will provide users with information, which is specific to the experiment they are planning.
For all scientists involved in the research process, the EDA is also extremely useful as a communication tool.
The interaction with the system and the build of the diagram emulate the initial fact-finding discussion a statistician might have with a researcher. Using EDA diagrams facilitates the process by helping the researcher identify independently much of the information which will be required by a statistician to provide expert advice, and present it in an explicit and standardised format.
A central feature of the EDA is the use of a formal, diagrammatic notation to describe experimental plans and analyses. This is an approach that has been adopted by many technical disciplines to improve communications. It allows the design of an experiment to be recorded clearly and unambiguously and EDA diagrams help convey experimental plans efficiently: for example, between students and their supervisors, or with colleagues and collaborators. These visual representations are far more explicit than the cursory text description traditionally included in grant applications, ethical review submissions or journal publications. Our goal is to integrate the EDA into the scientific process to facilitate better peer review of experimental plans.
The EDA is an NC3Rs initiative and was developed as part of a programme of work on experimental design. Previous work by the NC3Rs and others have provided evidence that many animal experiments published in the scientific literature contain flaws which compromise the validity of the findings.
The EDA complements the NC3Rs ARRIVE guidelines for reporting animal research.
The EDA was initially developed as an NC3Rs-led collaboration between an expert working group of in vivo researchers and statisticians, and a team of software designers who specialise in innovative solutions for the life sciences. Experts providing advice on the EDA have changed over time.
Current advisory group
|Dr Simon Bate||GlaxoSmithKline|
|Dr René Bernard||Charité Universitätsmedizin Berlin|
|Dr Yu-Mei Ruby Chang||Royal Veterinary College|
|Dr Darren Dahly||University College Cork|
|Dr Derek Fry||University of Manchester|
|Dr Natasha Karp||AstraZeneca|
|Dr Stanley Lazic||Prioris.ai|
|Professor Malcolm Macleod||University of Edinburgh|
|Dr Richard Preziosi||University of Plymouth|
|Professor Helene Richter||University of Münster|
|Professor Clare Stanford||University College London|
|Dr Tracey Weissgerber||Berlin Institute of Health at Charité Universitätsmedizin Berlin|
|Dr Manuel Berdoy||University of Oxford|
|Dr Robin Clark||Envigo*|
|Professor Innes Cuthill||University of Bristol|
|Professor Lawrence Moon||King's College London|
*indicates the expert was at this institution when advising on the EDA.
The ontology and grammar used in the diagrams were developed using an iterative approach and tested using a wide range of experimental plans from the published literature. Before it was launched, the EDA underwent two formal phases of user testing, where NC3Rs grant holders and other contacts were tasked with designing an experiment in the EDA and answering specific questions.
The development of the rule set, which underlies the feedback feature (Critique and Analysis Suggestion functionalities), was informed by workshops with the working group where EDA diagrams representing examples of flawed experimental designs were discussed and analysed. In particular, the group identified what information a statistician would need to have in order to offer the best advice on the design and what information the statistician would feedback to the researcher to help them identify the requested information, or improve the design. This information was then organised in a rule set, consisting of prompts (either asking for more information or providing advice) and triggers (which define specific situations where the prompt should be displayed). The rule set undergoes automated checks daily to ensure that further development of the system does not impact on the behaviour of existing rules that have been previously tested and implemented.
We intend to build on the EDA to support a wide range of experimental designs and ensure users benefit as much as possible from the system.
We would also like to encourage users to tell us about the quality of the feedback they obtain from the system. To provide feedback please contact us. This will support the process of continual improvement to which we are committed.
Citing the EDA
Publications citing the EDA should use the format: EDA (Experimental Design Assistant: RRID:SCR_017019, https://eda.nc3rs.org.uk)
The EDA was recently described in the following articles:
Percie du Sert N, Bamsey I, Bate ST, Berdoy M, Clark RA, Cuthill IC, Fry D, Karp NA, Macleod M, Moon L, Stanford SC, Lings B (2017) The Experimental Design Assistant. Nat Methods. doi: 10.1038/nmeth.4462. https://doi.org/10.1038/nmeth.4462
Percie du Sert N, Bamsey I, Bate ST, Berdoy M, Clark RA, Cuthill I, Fry D, Karp NA, Macleod M, Moon L, Stanford SC, Lings B (2017) The Experimental Design Assistant. PLoS Biol 15(9): e2003779. doi: 10.1371/journal.pbio.2003779 https://doi.org/10.1371/journal.pbio.2003779
Cressey D (2016) Web tool aims to reduce flaws in animal studies. Nature, 531:128. https://doi.org/10.1038/531128a