Machine Learning Model Predicts Parkinson’s with 79% Accuracy

parkinsons disease

In a scientific breakthrough, researchers trained an ML model to predict Parkinson’s disease to a high degree of accuracy.

Parkinson’s is one of the toughest diseases to deal with. With an increasing average lifespan all over the world, the disease is also increasing as it mainly affects older people. It causes the degeneration of brain, leading to slow movement, tremors, imbalance, lack of control over the motor nerves, and so on.

22 researchers published a paper in Nature on the identification of biomarkers to predict Parkinson’s disease up to 7 years before the onset of its symptoms. Using a machine learning model, the researchers were able to diagnose individuals that might develop Parkinson’s disease well in advance.

This is a major breakthrough in science. The model identified all Parkinson’s patients (that ended up getting the disease) with a 79% accuracy. And it did that in a record time of 7 years before the actual onset of the symptoms. The study utilized motor Parkinson’s patients, pre-motor individuals with REM disorders, and a healthy control group.

There have been similar breakthroughs in the past. The way these studies work is that they train a machine learning model on pre-existing data, then give it the details of people who are already diagnosed with the disease, without telling it so. The model’s job is to try and identify who will develop the disease. In this case, the model was 79% accurate, meaning it correctly predicted the onset of the disease in 79% of the individuals that actually have the disease now.

The obvious use of this research is to put the model to work (and improve its accuracy) over time. Hopefully, this (or a more researched variant of the model) will allow us to diagnose Parkinson’s disease in individuals who are at-risk more often. Early diagnosis can then pave the way for better treatment plans and expectations.

By Abhimanyu

Unwrapping the fast-evolving AI popular culture.