Scientists from the University of New South Wales (UNSW) in Sydney, in collaboration with researchers from Boston University, have developed a machine-learning tool that shows promise in predicting the onset of Parkinson’s disease before the appearance of symptoms. This breakthrough, described in a recent publication in the journal ACS Central Science, involves the analysis of biomarkers in patients’ bodily fluids using neural networks.
The UNSW research team focused on blood samples obtained from healthy individuals participating in the Spanish European Prospective Investigation into Cancer and Nutrition (EPIC) study. They specifically examined samples from 39 individuals who later developed Parkinson’s disease up to 15 years later. By utilizing their machine learning program, the scientists were able to identify unique combinations of metabolites, the chemical compounds produced when the body breaks down food, drugs, or chemicals, which could potentially serve as early warning signs or preventative measures for Parkinson’s.
The machine learning tool developed by UNSW chemists Diana Zhang and Associate Professor W. Alexander Donald, called CRANK-MS (Classification and Ranking Analysis using Neural network generates Knowledge from Mass Spectrometry), takes into account the associations between different metabolites, providing a more comprehensive analysis compared to conventional statistical approaches. Rather than reducing the number of chemical features, the tool incorporates all available data to generate predictions and identify the metabolites that play a key role in the disease.
Parkinson’s disease is currently diagnosed based on the observation of physical symptoms such as hand tremors. There are no blood or laboratory tests available for the non-genetic form of the disease. However, atypical symptoms like sleep disorders and apathy can manifest in individuals with Parkinson’s many years before the motor symptoms become apparent. CRANK-MS could potentially be used to assess the risk of developing Parkinson’s when these atypical symptoms first emerge.
While the results of this study are promising, Associate Professor Donald stresses the need for further validation using larger cohorts and diverse populations before the tool can be reliably implemented. In the limited cohort analyzed for this study, CRANK-MS demonstrated an accuracy of up to 96% in detecting Parkinson’s disease by analyzing blood chemicals.
The study also uncovered interesting findings regarding the metabolites of individuals who later developed Parkinson’s disease. For instance, lower concentrations of triterpenoids, known neuroprotectants found in foods like apples, olives, and tomatoes, were observed in the blood of these individuals. This raises the possibility that consuming these foods could naturally help protect against the development of Parkinson’s disease.
In conclusion, the development of CRANK-MS represents a significant advancement in the early detection and prediction of Parkinson’s disease. With further research and validation, this machine-learning tool has the potential to revolutionize the diagnosis and management of the disease, leading to more proactive and effective interventions. Additionally, the identification of key metabolites opens up new avenues for investigating preventive measures and exploring the role of diet in reducing the risk of Parkinson’s disease.