91直播

Stuart Wright 2024 Open Research Award winner

Last year I was honoured to be awarded the inaugural University of Manchester Open Research Award for our work on the MANC-RISK-SCREEN model. This model, which simulates different strategies for providing breast cancer screening to UK women, went through a published validation process, was published as a pre-print, and the model code was made publicly available through GitHub. At the time of last year's Open Research Conference, the model was being used by a PhD student in Aberdeen to evaluate the use of artificial intelligence in breast cancer screening.

Since that time, the open nature of the model has propelled our research and helped as to develop wide ranging collaborations. Our team were approached to join a team leading a large scale trial of the use of artificial intelligence in breast screening in the UK. We will adapt the model, working with and building on the work of the PhD student in Aberdeen, particularly around modelling the results of the trial in Scotland. In a different project, the model will be adapted by researchers at the University of Cambridge to evaluate the introduction of a breast cancer risk prediction service for younger women in the UK. We have also been approached to support a grant application to explore the impact of lowering the starting age of breast cancer screening in the UK. The code on GitHub has also been accessed by researchers in the Netherlands and Iran.

Our experience with this model has further cemented our desire to adopt a philosophy of open modelling in our future work. Not only has making the model open source provided more opportunities for collaboration and funding but each new project means different sets of eyes on the model code to identify errors and areas for improvement. The outputs of each project add additional components and capabilities to the model, extending is usefulness further. This provides multiple avenues for informing UK health policy and improving breast cancer screening for women.

We have also recently become consumers of Open Research as well as producers, attempting to build on existing open source models to develop screening models in cervical and lung cancer screening. For the cervical cancer model, we have been provided with existing code for a Human Papillomavirus infectious disease model and are exploring the detailed HPVsim model coded in Python (which is helping me to develop my own coding skills in the language!). For our lung cancer model, we hope to build on the work of colleagues at the University of Exeter whose open source ENaBL model informed the introduction of lung screening in the UK. Accessing these models helps us to speed up our research and will hopefully allow us to add components to the code which will benefits other researchers.

Hopefully this post has convinced you of the value of producing and using open source code in your research. If so, I鈥檇 like to end on a few tips. Even though you may be able to openly access code, it will still take a reasonable amount of effort to understand that code and adapt it for your own purposes. We would therefore highly recommend contacting the original developers to see if they can provide some advice to get you started. Hopefully (as in our case) they will be happy to provide a bit of an introduction to the code. If you need more detailed input or need to conduct your research quickly then you could also discuss the opportunity for paper authorship in return for their input. For major pieces of work where you need a lot of advice you should also consider including model developers as co-applicants on grants. Finally, even if you do decide to adapt the code yourself, its still useful for code developers to know that their model is being used so they can highlight their impact so don鈥檛 forget to drop them an email!