May 19, 2023 14:00:00

An AI tool will be developed to examine the risk of developing Parkinson's disease ``before symptoms appear'' in a blood test

Parkinson's disease is a neurodegenerative disease that presents with movement disorders such as hand tremors and difficulty walking.It is a serious disease that can lead to wheelchairs and bedridden lifestyles as the symptoms progress. An AI tool `` CRANK-MS '' that finds signs of Parkinson's disease from blood tests before such clear symptoms of Parkinson's disease appear was developed by research teams at the University of New South Wales and Boston University .

Interpretable Machine Learning on Metabolomics Data Reveals Biomarkers for Parkinson's Disease | ACS Central Science
https://doi.org/10.1021/acscentsci.2c01468

Scientists develop AI tool to predict Parkinson's disease onset | UNSW Newsroom
https://newsroom.unsw.edu.au/news/science-tech/scientists-develop-ai-tool-predict-parkinsons-disease-onset

An AI Was Trained to Detect Parkinson's Years Before Symptoms Appeared : ScienceAlert
https://www.sciencealert.com/an-ai-was-trained-to-detect-parkinsons-years-before-symptoms-appeared

A research team led by University of New South Wales researchers Diana Chang and Associate Professor Alexander Donald uses neural networks to perform mass spectrometry-based classification and We have developed an AI tool 'CRANK-MS' that performs ranking analysis.

“The most common way to analyze metabolomics data is through a statistical approach,” Chan said. usually look for correlations about a specific molecule, but here we adopted machine learning, considering that metabolites may be related to other metabolites. We are using the computational power of AI to elucidate what is happening in the presence of metabolites.'

The research team used plasma samples collected in Spain as part of the European Prospective Study of Nutrition and Cancer (EPIC) to study 39 individuals who developed Parkinson's disease within 15 years of sample collection and Parkinson's disease. A control group of 39 individuals who did not develop disease were identified. Each plasma sample was analyzed using 'CRANK-MS'.

“Typically, researchers using machine learning to correlate metabolites to disease would reduce the number of chemical features before feeding them into the algorithm,” said Donald. Instead, we feed all the information into CRANK-MS, from which we can get the predictions of the model and identify which metabolites have the most influence on the predictions, all in one step. This means that even if there are metabolites that might be missed by traditional approaches, they can now be picked up.'

As a result of the study, it was found that 'CRANK-MS' can detect people who develop Parkinson's disease later in life with 96% accuracy based on the chemical substances present in the plasma sample. By identifying the risk before clear symptoms of Parkinson's disease appear, it is expected to prevent the onset or delay the progression.

We also found several features in the plasma of people who developed Parkinson's disease later in life. One of them is that the concentration of compounds called 'triterpenoids', in which

triterpenes are modified, is low in the plasma of people who later develop Parkinson's disease. Triterpenoids play the role of neuroprotective agents that regulate oxidative stress, and are abundant in foods such as apples, olives, and tomatoes. Future research could investigate whether eating foods high in triterpenoids can reduce the risk of developing Parkinson's disease.

In addition, we found that the plasma of people with Parkinson's disease contained more chemicals called polyfluoroalkyl compounds (PFAS) that do not exist in nature. PFAS is also called an 'eternal chemical substance' because it is extremely difficult to decompose, and it is a substance that is regarded as a problem because it is harmful to the human body. 'There is evidence to suggest that PFAS is involved in Parkinson's disease, but more data is needed to be 100% sure,' said Donald.

At the time of writing, Parkinson's disease is diagnosed by physical symptoms such as resting hand tremors, and there are no blood or laboratory tests to diagnose non-genetic cases. However, some atypical symptoms, such as sleep disturbances and apathy toward things, may appear decades before motor symptoms appear. Therefore, by using 'CRANK-MS' in patients with atypical symptoms, it is expected to identify the risk of developing Parkinson's disease in the future.

Mr. Donald emphasizes that because this experiment is only a small scale, a larger verification study is needed before expanding the use of 'CRANK-MS'. Still, the results of this time are very promising and interesting.

“The use of CRANK-MS to detect Parkinson’s disease is just one example of how AI can improve disease diagnosis and monitoring,” Chan said. It can be easily applied to other diseases and identify new biomarkers of interest, ”he argued that ``CRANK-MS'' may be used for early diagnosis of other diseases.