A team of scientists from UNSW Sydney and Boston University may have found a way to spot Parkinson’s disease long before the typical symptoms show up—and it’s all thanks to artificial intelligence.
They’ve developed a powerful AI tool called CRANK-MS that can analyze blood samples and predict the likelihood of developing Parkinson’s with up to 96% accuracy. That’s a big deal because currently, there are no blood tests available to diagnose Parkinson’s, especially the non-genetic kind.
What’s CRANK-MS?
CRANK-MS stands for Classification and Ranking Analysis using Neural network generates Knowledge from Mass Spectrometry. It might sound complicated, but think of it as a super-smart program that learns from huge sets of chemical data found in our blood—called metabolites—to spot patterns linked to disease.
Instead of looking at one molecule at a time (which is what traditional methods do), CRANK-MS looks at all the data together and picks out combinations of chemicals that may signal the early stages of Parkinson’s. Even more impressively, it doesn’t filter or trim the data before analyzing—it takes in everything, which helps it find subtle clues that might otherwise be missed.
How Did They Test It?
The researchers used blood samples from a long-term European health study. They compared 39 people who later developed Parkinson’s with 39 who didn’t. CRANK-MS analyzed the samples and identified specific patterns in the blood that were more common in the people who eventually developed the disease.
One interesting finding? People who went on to develop Parkinson’s had lower levels of triterpenoids—compounds known to protect brain cells. These are found in everyday foods like apples, tomatoes, and olives, which opens the door to future research about how diet might influence Parkinson’s risk.
Another eyebrow-raising discovery was a possible link between PFAS (industrial chemicals) and increased Parkinson’s risk—though more research is needed to confirm this.
What This Means
Right now, Parkinson’s is diagnosed by observing visible symptoms, like tremors or stiffness. But signs like sleep issues or apathy can appear decades earlier. This tool could give doctors a new way to detect the disease much sooner—possibly even before serious symptoms begin.
It’s still early days, and the team is clear that larger, global studies are needed before CRANK-MS can be used in real-world clinics. But this early success is incredibly promising.
And the Best Part?
CRANK-MS is free and open to all researchers. It runs on a regular laptop and takes about 10 minutes to deliver results, making it easy to use and widely accessible.
While the researchers are focused on Parkinson’s for now, the tool can be applied to other diseases too. That means CRANK-MS could one day help in early detection of many health conditions—all by studying what’s already in your blood.
Comments