Alan Turing methods could form basis of better early diagnosis tests for cancer

Mathematical techniques developed by computer pioneer Alan Turing could form the basis of a more reliable test for the early diagnosis cancer and other diseases, according to new research.

Work by the Second World War cryptographer – who famously cracked the German ‘Enigma’ code, arguably shortening the conflict – could help doctors in the emerging field of ‘precision medicine’, which aims to personalise treatments so that patients have a better chance of recovery.

Turing’s so-called ‘weights of evidence’ technique is superior to the existing C-statistic method – which has been used by researchers since the 1980s as the statistical model used to quantify the performance of a diagnostic test.

However, the C-statistic does not indicate how the diagnostic test would perform when used to ‘stratify people by risk’, Professor Paul McKeigue, of the University of Edinburgh’s Usher Institute of Population Health Sciences and Informatics, argues.

In a blog post attached to an article published in the Statistical Methods in Medical Research journal, he writes: “I propose an alternative approach based on estimating the weight of evidence. The background to this work lies in unpublished studies of Alan Turing, who in 1941 at Bletchley Park investigated the distribution of weights of evidence to decide the best strategy for breaking the ENIGMA code. Turing discovered some key properties of this distribution, which were extended in 1968 by his former assistant Jack Good, by then one of the most influential Bayesian statisticians of the 20th century.”

Turing’s method could be used to address two major challenges – how likely a person is to develop a disease, and whether or not a drug will work for them.

The development could improve on existing statistical tools that can assess the accuracy of diagnostic tests. They are unable currently to gauge how useful a test might be in determining a particular individual’s risk of developing a disease.

Researchers say existing methods are not useful for evaluating the likely impact of a new test compared with an old one. Neither are they effective at predicting how often the test will give wrong results in practice.

Working at Bletchley Park in 1941, Turing devised his method used to break the German forces’ Enigma code; the story was told in the 2014 Hollywood blockbuster The Imitation Game, starring Benedict Cumberbatch.

Turing worked out how the weight of evidence was expected to vary over repeated experiments.  His ideas were developed further in 1968 and published by his former assistant Jack Good.

Professor Paul McKeigue, of the University’s Usher Institute of Population Health Sciences and Informatics, shows that the same principle of how the weight of evidence varies can be applied to evaluate the diagnostic tests used for personalised treatments. In this way, the performance of a test can be quantified.

He said “Most existing diagnostic tests for identifying people at high risk of cancer or heart disease  do not come anywhere near the standards we could hope to see. The new era of precision medicine is emerging, and this method should make it easier for researchers and regulatory agencies to decide when a new diagnostic test should be used.”


 

Join FutureScot’s Digital Health & Care conference on November 27 to learn how healthcare data is being used to deliver innovative new services.