Artificial intelligence can be trained to detect Parkinson's disease years before symptoms appear

Scientists are not only looking for a cure for Parkinson's disease, they are also busy looking for better ways to detect it early and halt its progression.

Above: The neural network is built on a weighted node layer.

Now, a new tool that can run quickly on a standard laptop uses artificial intelligence to detect potential signs of the disease years before symptoms such as tremors and slowed movements appear. It's called "CRANK-MS": classification and ranking analysis using neural networks to generate knowledge from mass spectrometry.

The tool uses layers of trained nodes that simulate the human brain to look for specific compounds (metabolites) in the blood, identifying patterns that may predict the presence of disease or prevent it.

To figure out which metabolites are more important for the disease compared to a control group, researchers often look at metabolites involving specific molecules, said Diana Zhang, a chemist at the University of New South Wales in Australia. Correlation. But here, we consider that metabolites may be related to other metabolites -- that's where machine learning comes in. For hundreds or thousands of metabolites, we use computational power to understand what's going on. ”

The research team used plasma samples that were part of the Spanish European Prospective Study on Nutrition and Cancer. The team looked at 39 patients who developed Parkinson's disease within 15 years of participating in the study and compared the metabolite mixture with a control group of 39 patients who did not develop Parkinson's disease. Several patterns were identified that were considered potentially important.

These metabolites are produced when the body breaks down food, drugs or chemicals. For example, the team noticed that people with Parkinson's disease tended to have lower blood levels of triterpenes, which deal with stress in the body at the cellular level and are found in foods like apples, olives and tomatoes.

Researchers also noticed the presence of polyfluoroalkyl substances (PFAS) in people who later developed Parkinson's disease. This may be related to high exposure to industrial chemicals, but larger studies involving more patients are needed to know for sure.

Although the study was relatively small, CRANK-MS was able to detect the risk of Parkinson's disease with 96% accuracy. This is partly due to the volume and breadth of data that is fed into the system from the start, without the need for manual simplification or filtering.

Above: Blood analysis can be used to assess the risk of Parkinson’s disease.

William Donald, a chemist at the University of New South Wales, said: "Here, we input all the information into CRANK-MS, without any data reduction at the beginning. From this, we can get the model predictions, and determine which metabolites are most driving the predictions, all in one step. This means that if there are metabolites that may have been missed by traditional methods, we can now extract those."

Other scientists can also use CRANK-MS. This means more diseases can be detected through blood samples.

The researchers now hope to see their system tested on larger cohorts in more parts of the world to see if the AI ​​analysis works in Parkinson's disease - but not in terms of analyzing metabolites in the blood. , early results are promising.

Chemist William Donald said: "First, the accuracy of predicting Parkinson's disease before clinical diagnosis is very high. Second, this machine learning method allows us to identify chemical markers that accurately predict the future The aspect of who will develop Parkinson's disease is most important. Third, some of the chemical markers that most accurately predict Parkinson's disease have been linked to Parkinson's disease by others in previous cell-based analyses, but Not in humans.”

The study was published in the journal ACS Central Science.

If friends like it, please pay attention to "Knowing the New"!

The above is the detailed content of Artificial intelligence can be trained to detect Parkinson's disease years before symptoms appear. For more information, please follow other related articles on the PHP Chinese website!