For years, diagnosing Parkinson’s disease has relied on a time-consuming series of neurological assessments, physical exams and careful clinical observation.
Now, scientists in India say a far simpler approach may help identify the condition with as much as 99 percent accuracy: a basic drawing test.
Parkinson’s disease is a severe neurological disorder in which damaged neurons trigger gradually worsening tremors, stiffness and movement difficulties, ultimately threatening a patient’s independence.
About one million Americans are living with Parkinson’s, and cases are believed to be increasing in the US, a trend some experts have linked to pollution, pesticide exposure and smoking.
For the research, the team examined data from an earlier study involving 66 participants, 31 of whom had Parkinson’s, who were asked to carry out two separate drawing tasks.
During the tests, volunteers traced spiral patterns and meanders — continuous lines with angular shapes — while using a biometric pen designed to record subtle hand movements.
The results showed that people with Parkinson’s had far greater difficulty following the lines accurately than participants without the condition.
In the latest analysis, researchers used that information to train a detection model, which they say could help screen for Parkinson’s disease through drawing-based movement patterns.
Researchers say that a handwriting trait could be used to detect Parkinson’s disease
In Parkinson’s, the breakdown of neurons can cause symptoms including tremors – movements outside the person’s control – which may leave sufferers unable to hold a pen steady.
Nearly all Parkinson’s patients experience tremors, which may emerge early in the disease or during its later stages.
Other conditions can also cause tremors including hyperthyroidism, low blood sugar, certain medications and withdrawal from substances including alcohol.
The team said in their study: ‘Handwritten images provide spatial characteristics of stroke irregularities, tremor-induced distortions and shape deviations.
‘In contrast, sensor-based handwriting signals capture motor behavior, which includes velocity fluctuations, pressure inconsistencies and coordination.’
In the study, published in the journal Discover Computing, researchers fed the images and data from hand movements into different AI systems.
Each model then evaluated spatial irregularities and differences in motor control and hand coordination between those who did and did not have the condition.
The data from each was then processed into an algorithm called SNAKE, which was then used to re-evaluate each drawing and determine which participants did or did not have Parkinson’s.
Shown above are the spirals, top row, and meanders, bottom row, that were used to test for Parkinson’s. The two drawings on the right are by Parkinson’s patients.
The above shows a meander drawn by a person who does not have Parkinson’s disease (left) and who has Parkinson’s
The above shows a spiral drawn by a person who does not have Parkinson’s disease (left) and does have Parkinson’s
According to the researchers, this algorithm correctly diagnosed Parkinson’s using the meander drawings in 98.95 percent of cases.
When analyzing spatial patterns, it correctly detected Parkinson’s in 97.7 percent of cases.
The researchers said the test could be a less invasive way to diagnose Parkinson’s. It was not clear whether it would also help to detect the disease in the early stages.
It is not clear whether the algorithm may now be used by doctors to help them confirm a Parkinson’s disease diagnosis.
The dataset was small, including 66 participants, and the algorithm was not evaluated using a new group of participants or new drawings.
The researchers, from Siksha ‘O’ Anusandhan University, concluded: ‘This study proposed a multimodal handwriting-based framework for Parkinson’s disease detection.’
