skip to primary navigation skip to content

Predict websites

Winton Centre for Risk and Evidence Communication

Project - Predict websites

predict-breast.png

Predict:Breast Cancer is an online tool that allows clinicians to compare the potential benefits of different treatment options for individual women post-breast cancer surgery. The results are based on an algorithm designed by Prof Paul Pharaoh and colleagues at the University of Cambridge's Cancer Epidemiology Unit, and they originally designed the tool which is hosted by Public Health England. It is the world’s most popular such decision tool in 2018: when we launched our new user interface it was already being used around 20,000 times per month. Now it is being used around 30,000 times per month.

We worked with clinicians, past patients and ‘ordinary people’ to develop an entirely new interface to the existing algorithmically-driven site. We are now preparing that work for publication so as to be able to share our experience with others.

As a result of our work, Predict:Breast Cancer is now suitable for use directly with patients during their appointments, and presents the information on the likely benefits to a particular individual patient of different treatment options in a range of formats. The default is a simple table, giving busy clinicians the quick reference to the bare numbers that they need when working fast. Other tabs then give visualisations of those numbers as curves, bar charts, icon arrays and text.

We introduced the option of seeing the uncertainty around the benefits (our work on the communication of uncertainty suggests that this does not make it harder for people to understand the information or make decisions based on it).

We are now carrying out more work on how people understand the numbers when they are presented in the different formats (which will be published when we have completed it), and are working with the team behind the algorithm to add functionality. Later in 2019 we hope to add:

  • the effects of extending endocrine therapy from 5 years to 10 years post surgery
  • the potential harms of the treatment options (we are currently undertaking meta-analyses to quantify these)
  • the effects of the different therapy options on recurrence of breast cancer (not just mortality)

We are also working on providing translations of the site into several languages other than English.

Colleagues Prof Sarah Darby and team at the University of Oxford are working on adding radiotherapy as a potential treatment option.




Predict: Prostate Cancer

Predict:Prostate Cancer is a new sister site for Predict:Breast Cancer, launched in February 2019 for men diagnosed with non meta-static prostate cancer. The algorithm behind it was developed by Vincent Gnanapragasam and David Thurtle at the University of Cambridge's Department of Urology. It is already being used over 500 times per month.

Predict:Prostate Cancer uses the same design as Predict:Breast Cancer, except that it already includes the potential harms of the treatment options. We are working with the team to add functionality and develop a future version for men with metastatic prostate cancer, and also hope to be able to provide translations into languages other than English.

We are keen to help teams develop algorithms for other cancers which might be able to join the growing Predict stable of tools.