New research published in Scientific Reports demonstrates that machine learning methods can predict with a 35% improvement in accuracy whether a cystic fibrosis (CF) patient should be referred for a lung transplant, in comparison to existing statistical methods. It is the first machine learning study to make use of a dataset representing 99% of CF patients living in the UK, the CF Registry.

The research, led by Professor Mihaela van der Schaar (The Alan Turing Institute, University of Oxford) has been generated through a partnership between The Alan Turing Institute and the Cystic Fibrosis Trust, a charity with the mission to beat cystic fibrosis for good.

Cystic fibrosis is a genetic condition that affects more than 10,000 people in the UK and causes a wide range of challenging symptoms which affect the entire body. Lung transplants may be recommended as the last treatment option, when others are no longer having an impact. However, the procedure comes with serious risk of post-transplant complications, and the availability of lung donors is scarce in comparison with demand.

In order to explore whether the lung transplant process could be improved through machine learning methods, the Cystic Fibrosis Trust supplied lead author Mihaela van der Schaar with access to an anonymised extract of UK CF Registry data. Using the dataset, Mihaela and her co-author, PhD student Ahmed Alaa (UCLA), developed an algorithmic framework that leverages machine learning to automate the process of constructing a prognostic model for CF patients – the point at which a clinician assesses a patient and calculates the risks of taking a certain pathway against the projected benefits. The new algorithmic model, called AutoPrognosis, is capable of achieving a positive predictive value of 65%. Existing practice results in only 48% of individuals being correctly referred, so AutoPrognosis yields a 35% increase in accuracy overall in comparison to current methods.

The authors and experts from the Cystic Fibrosis Trust believe that this research could be used to significantly augment clinicians’ decision-making processes. Use of machine learning to develop improved prognostic tools, provides additional information for clinicians and patients to assist in discussing future treatment options.  Furthermore, the machine learning methods developed in this study could be applied for other diseases in future, for example heart attacks or cancer diagnosis.

Dr Janet Allen, Director of Strategic Innovation at the Cystic Fibrosis Trust, commented:

“We are delighted to be working with colleagues at The Alan Turing Institute on this project. By working collaboratively in this way, we will make significant steps in understanding cystic fibrosis and improving the lives of those affected by it.

For doctors and clinical teams, making decisions with their patients on whether they should be considered for a lung transplant is difficult. Accurate methods to help make that call are vital. This research would not have been possible without the generous consent of people with CF to share their data in the UK CF registry.”

Mihaela van der Schaar, Fellow of The Alan Turing Institute and Professor in the Department of Engineering Science at the University of Oxford, said:

“While machine learning has proven successful in making predictions in a clinical setting, its deployment in practice has been limited. The outcomes of our research with the Cystic Fibrosis Trust demonstrate that with the right in-depth expertise, anonymised data from a large population, and input from clinicians, we can create algorithmic methods to support clinicians in their day-to-day decision-making.

I am grateful to the Trust for their support and advice and for insights from patients with cystic fibrosis we have worked with in the course of this study. We look forward to continuing to work together to ensure that our work is useful for stakeholders such as patients, families of patients, clinicians and policymakers, for instance.”

Professor Andres Floto, University of Cambridge and CF Physician, Royal Papworth Hospital, added:

“This remarkable work elegantly demonstrates that machine learning is now ready for the clinic, will have an immediate impact on how we think about who to refer for transplantation, and could have tremendous benefits for individuals with CF. We hope to now build on these exciting developments to discover more about the biology of CF and how best to manage it.”

Media Inquiries

Beth Wood 
Press and Communications Manager
The Alan Turing Institute
[email protected] +44 (0)20 3862 3390

Cystic Fibrosis Trust Press Office
(+44) 0203 7952 193
[email protected]

Notes to editors

  1. Mihaela van der Schaar is a Turing Fellow at The Alan Turing Institute in London and is MAN Professor of Quantitative Finance in the Oxford - MAN Institute (OMI) and the Department of Engineering Science at Oxford, and Fellow of Christ Church College. Ahmed Alaa is van der Schaar’s PhD student (UCLA) and is working on similar methods for his thesis. 
  2. The method was highlighted at ICML in July 2018 and is using cardiovascular datasets. For more, see A. M. Alaa, M. van der Schaar, "AutoPrognosis: Automated Clinical Prognostic Modeling via Bayesian Optimization with Structured Kernel Learning," ICML, 2018.
  3. A paper van der Schaar authored with Oxford PhD student Bryan Lim also received Best Paper Award in a workshop dedicated to medicine at IJCAI BOOM workshop (held in conjunction with ICML): B. Lim, M. van der Schaar, "Disease-Atlas: Navigating Disease Trajectories using Deep Learning," IJCAI-BOOM, 2018.
  4. A short clinical abstract based on this work was accepted for oral presentation at the North American Cystic Fibrosis Conference and one of the clinical collaborators, Dr. Andres Floto, will present it in October 2018. The work was also presented in June 2018 at the European Cystic Fibrosis Conference by another clinical collaborator, Dr Thomas Daniels.
  5. Top image is copyright of Cystic Fibrosis Trust.

About Cystic Fibrosis

  • Cystic fibrosis is an inherited disease caused by a faulty gene. This gene controls the movement of salt and water in and out of your cells, so the lungs and digestive system become clogged with mucus, making it hard to breathe and digest food
  • Cystic Fibrosis is a life shortening genetic condition – half of those that died with cystic fibrosis in the past 12 months were under the age of 31
  • There are over 10,400 people with cystic fibrosis living in the UK and the population is growing every year
  • Two million people in the UK are carrying the faulty gene without realising it. If two carriers have children, there’s a one in four chance their child will have the condition, which slowly destroys the lungs and digestive system
  • People with cystic fibrosis often look perfectly healthy. But it’s a lifelong challenge involving a vast daily intake of drugs, time-consuming physiotherapy and isolation from others with the condition. It places a huge burden on those around them and the condition can critically escalate at any moment
  • Half of people with cystic fibrosis alive today are expected to live into their forties, thanks to earlier diagnosis and ongoing developments in care and treatments
  • Visit www.cysticfibrosis.org.uk to find out more about cystic fibrosis and the work of the Cystic Fibrosis Trust.