Researchers have developed an AI-enhanced blood test that may predict the onset of Parkinson’s disease as much as seven years before symptoms manifest.
The study, led by scientists from University College London (UCL) and the University Medical Center Goettingen in Germany and published in Nature Communications, could unlock early, targeted treatments to slow the progression of this debilitating neurodegenerative disorder.
Parkinson’s disease is a rising global health concern, affecting nearly 10 million people worldwide, including over 1 million individuals within the US and 150,000 people within the UK.
The disease is characterised by symptoms resembling tremors, movement difficulties, muscle stiffness, balance issues, memory problems, dizziness, and nerve pain. Symptoms arise on account of the death of nerve cells within the “substantia nigra,” an element of the brain liable for controlling movement.
Currently, there are not any treatments available to halt or reverse the disease progression, and most patients are diagnosed only after symptoms have already developed.
Dr Jenny Hällqvist from UCL, a study co-author, explained the necessity for proactive measures, saying, “People are diagnosed when neurons are already lost. We must protect those neurons, not wait until they’re gone.”
Research News 📣
Parkinson’s UK funded research shows promise for a blood
test that would discover Parkinson’s before movement symptoms occur.
Read the total story and what this implies for individuals with Parkinson’s 👉🏽 https://t.co/2LwhkHRXbf pic.twitter.com/yxTZHRSgJQ
Here’s how the study worked:
- Identifying potential biomarkers: The study began by analyzing blood samples from recently diagnosed Parkinson’s patients and healthy controls using advanced mass spectrometry techniques. This allowed the researchers to discover 47 proteins that were expressed otherwise between the 2 groups.
- Developing a targeted blood test: From their initial evaluation, the team developed a targeted blood test to measure the degrees of 121 specific proteins.
- Validating the test: The researchers then applied the targeted test to blood samples from an independent group of Parkinson’s patients, healthy controls, individuals with other neurological disorders, and patients with isolated REM sleep behavior disorder (iRBD), a known risk factor for Parkinson’s. This confirmed that 23 of the measured proteins significantly differed between Parkinson’s patients and healthy controls.
- Applying machine learning: The data from the validation step was used to coach machine learning models to differentiate between Parkinson’s disease and healthy controls based on protein levels. A model using just eight of the proteins was capable of accurately classify Parkinson’s and healthy samples with 100% accuracy. Impressively, the model also predicted that 79% of the iRBD samples were Parkinson’s-like, suggesting that the test could discover individuals at high risk of developing the disease.
- Results: To further confirm the findings, the researchers refined the test and applied it to a separate group of 54 iRBD patients who had provided 146 blood samples over time. The machine learning models predicted that 70-79% of those samples were Parkinson’s-like, with a few of these predictions being made as much as 7 years before the individuals developed Parkinson’s symptoms.
The headline here is that the AI-enhanced blood test could predict Parkinson’s disease with as much as 79% accuracy as much as seven years before symptoms surface.
The team now plans to simplify the test further, allowing patients to easily mail a drop of blood on a card to the lab for evaluation.
Professor David Dexter, research director at Parkinson’s UK, a charity that helped fund the study, lauded the findings, stating, “The findings add to an exciting flurry of recent activity towards finding an easy solution to test for and measure Parkinson’s.” He also suggested that the test may have the ability to distinguish between Parkinson’s and other similar conditions.
While larger trials are crucial to validate the accuracy and reliability of this AI-enhanced blood test, it represents a large step forward in the hunt for early diagnosis of Parkinson’s disease.
It’s not the primary time AI has been deployed to discover and diagnose Parkinson’s either. Not way back, researchers developed an eye fixed test that would similarly discover the disease before symptoms develop.
Google’s longstanding AlphaFold project shows promise in discovering precisely how diseases like Parkinson’s develop, and the University of Cambridge researchers developed a model for locating Parkinson’s drugs that were 1000 times faster than conventional methods.
