Scientists from UNSW Sydney with collaborators at Boston University have developed a tool that shows early promise in detecting the onset of Parkinson’s disease years before the first symptoms appear.

The research, which uses neural networks to analyse biomarkers in patients’ bodily fluids, was published in the journal ACS Central Science, and predicted the onset of Parkinson’s disease with 96 per cent accuracy – and up to 15 years before symptoms appeared.

The researchers from UNSW School of Chemistry examined blood samples taken from healthy individuals gathered by the Spanish European Prospective Investigation into Cancer and Nutrition (EPIC). Focusing on 39 patients who developed Parkinson’s up to 15 years later, the team ran their machine learning program over datasets containing extensive information about metabolites – the chemical compounds that the body creates when breaking down food, drugs or chemicals.

UNSW researcher Diana Zhang explains that, along with Associate Professor W. Alexander Donald, she developed a machine learning tool called CRANK-MS, which stands for Classification and Ranking Analysis using Neural network generates Knowledge from Mass Spectrometry.  She told the Sydney Morning Herald that she first became interested in finding a diagnosis for Parkinson’s after hearing the story of Joy Milne, who claimed to be able to diagnose Parkinson’s by smell – years before symptoms appear. When scientists tested her claims by giving her T-shirts worn by people with and without Parkinson’s, Milne correctly assessed 11 out of the 12 cases.

What this means for people with Parkinson’s disease

Currently, a diagnosis for Parkinson’s relies on observing physical symptoms, but non-motor symptoms can present in people with Parkinson’s years or even decades before motor symptoms such as tremor appear. CRANK-MS could be used at the first sign of atypical symptoms such as sleep disorder to assess the risk of developing Parkinson’s in the future. Diana Zhang told the Sydney Morning Herald, “Basically, what it’s doing is going through all these different combinations possible to see which gives you the best prediction [of disease]”.

Associate Professor W. Alexander Donald says, “This study is interesting at multiple levels. First, the accuracy is very high for predicting Parkinson’s disease in advance of clinical diagnosis. Second, this machine learning approach enabled us to identify chemical markers that are the most important in accurately predicting who will develop Parkinson’s disease in the future. Third, some of the chemical markers that drive accurate prediction the most have been previously implicated by others to Parkinson’s disease in cell-based assays but not in humans.”

CRANK-MS is a tool that is publicly available to any researchers who would like to use machine learning for disease diagnosis using metabolomics data. Diana Zhang says the model was built so that it’s fit for purpose and user-friendly – on average, results can be generated in less than 10 minutes on a conventional laptop.