Researchers in India have developed a drawing test that can detect Parkinson's disease with up to 99 percent accuracy, according to a study published in the journal Discover Computing. This method aims to offer a less invasive alternative to the traditional diagnostic process, which often involves detailed neurological and physical examinations.
Parkinson's disease is a progressive neurological disorder that damages neurons, leading to tremors and worsening movement problems. It affects an estimated one million Americans, with some experts attributing an increase in cases to environmental factors like pollution and pesticides, as well as smoking.
The new approach utilizes a biometric pen to track hand movements while participants complete drawing tasks, such as tracing spirals and meanders. These drawings, along with the pen's data, were analyzed by artificial intelligence systems. The AI evaluated spatial irregularities and motor control differences between individuals with and without Parkinson's.
An algorithm called SNAKE was developed using this data. When applied to the meander drawings, the algorithm correctly identified Parkinson's in 98.95 percent of cases. For spatial patterns analyzed from the drawings, the accuracy rate was 97.7 percent. Researchers noted that handwritten images provide insights into stroke irregularities and tremor-induced distortions, while sensor-based signals capture velocity fluctuations, pressure inconsistencies, and coordination issues.
Tremors are a common symptom of Parkinson's, occurring when neuron damage affects control over movements. While tremors can be an early or late-stage symptom, they can also be caused by other conditions, including hyperthyroidism, low blood sugar, certain medications, and substance withdrawal.
The study's findings are based on a dataset of 66 participants, 31 of whom had Parkinson's. The researchers, from Siksha 'O' Anusandhan University, concluded that their multimodal handwriting-based framework shows promise for Parkinson's detection. However, the study did not evaluate the algorithm on a new set of participants, and it remains unclear if this method can detect the disease in its early stages or if it will be adopted by clinicians for diagnostic confirmation.