News

Life saving mathematics



14 September 2015

Dr Clio Cresswell, the University of Sydney's mathematics spokesperson
Dr Clio Cresswell, the University of Sydney's mathematics spokesperson

"Sometimes a sample space of one is big enough," explains Associate Professor Samuel Müller, a statistician in the School of Mathematics and Statistics at the University of Sydney. He then goes on to give the example of Professor Barry Marshall, the Western Australian physician who infected himself by drinking a solution containing the bacterium Helicobacter pylori to prove once and for all, after not being listened to for so long, that this bacterium was the cause of most stomach ulcers. This was indeed enough proof. And Professor Marshall went on to win the Nobel Prize in Physiology or Medicine in 2005. Not for any of the temporary self-harm part, you understand!

Our Associate Professor Müller is fascinated by sample spaces and how to extract information from them. He looks for patterns in data from a variety of disciplines, such as chemistry, biology and economics. Some of his recent work uses medical information with an analysis of melanoma data. You have to get very specific and precise, as these problems are so complex. There's no one single magic technique that works across everything.

Melanoma is a significant health problem worldwide that accounts for 0.1% of total global mortality and its markedly variable survival outcomes makes its clinical management particularly challenging. It would be helpful if we could at least clearly identify sluggish versus aggressive versions of the disease.

Today we not only have access to clinical data such as the patient's age, gender and birth place, and pathology results, but we also have access to large amounts of detail about gene, protein and microRNA expression too. So not just information about your genes, but also how your body is working with them. Clinical and pathology data can deliver up to about a hundred pieces of information. The other cellular data typically runs from thousands to tens of thousands of pieces of information. So the questions become: "How much data do I really need?" and "How can I best assimilate it to create a working model for us to understand the progression of the disease and therefore yield better treatment options and outcomes?" Maybe one piece of information is enough here too?

Not quite. But you see my point. Recently Associate Professor Müller worked with another statistician here Associate Professor Jean Yang, with their joint PhD student Kaushala Jayawardana, as well as other researchers from the Sydney Medical School, Melanoma Institute Australia, Discipline of Surgery, Discipline of Pathology as well as Tissue Pathology and Diagnostic Oncology, and found the following: clinico-pathologic information was not out-performed in terms of prognostic utility by any of the other large amounts of cellular data; and what performed best was a particular combination of clinico-pathologic variables and mRNA expression data. Sophisticated statistics bringing together researchers from across an array of expertise to significantly help people's lives.

This is why Associate Professor Müller fell in love with statistics, he says. It's the best place where he could use the latest mathematics and be relevant across such a diversity of fields, where good communication skills are equally vital. You need to explain your intricate and unique findings clearly to those whom you are collaborating with, and also those who will eventually be using your work. For him, statistics is that one place he found to use so many skills requiring such different parts of the brain. We only have a sample size of one of those each, so let's make the most of it!