A simple drawing test, enhanced by artificial intelligence, may offer a highly accurate and less invasive method for detecting Parkinson's disease, according to research from India. This approach could potentially streamline the diagnostic process, which traditionally relies on extensive neurological and physical examinations.
Parkinson's disease is a neurodegenerative disorder characterized by the progressive damage of neurons, leading to symptoms such as tremors, rigidity, and difficulties with movement that can significantly impact a patient's independence. One million Americans are affected by the condition, with some experts attributing its rising prevalence to factors like pollution, pesticides, and smoking.
The new method leverages a study involving 66 participants, 31 of whom had Parkinson's. Participants were asked to draw spirals and meanders (angular, continuous lines) while holding a biometric pen that recorded their hand movements. Analysis showed that individuals with Parkinson's struggled more to trace these lines compared to those without the condition.
Researchers then used this data to train an AI model. This model analyzed both the visual characteristics of the drawings and the data from the pen's sensors to identify patterns indicative of Parkinson's. The algorithm, named SNAKE, processed spatial irregularities, tremor-induced distortions, and motor control inconsistencies.
Diagnostic Accuracy
When applied to the meander drawings, the SNAKE algorithm achieved an accuracy rate of 98.95 percent in diagnosing Parkinson's. The analysis of spatial patterns from both drawing types yielded a detection rate of 97.7 percent.
IMAGE2
IMAGE4
IMAGE5
Handwriting and drawing irregularities, such as tremors, inconsistent pressure, and velocity fluctuations, are common symptoms of Parkinson's disease due to the breakdown of neurons. While tremors are a hallmark symptom, they can also be caused by other medical conditions. The researchers noted that handwritten images capture spatial characteristics of stroke irregularities, while sensor-based signals record motor behaviors.
IMAGE3
While the study, published in Discover Computing, demonstrates the potential of this multimodal, handwriting-based framework, the researchers acknowledge limitations. The dataset was small, and the algorithm has not yet been tested on a new set of participants or drawings. It remains unclear if this test can detect the disease in its early stages or if it will be adopted by clinicians for diagnostic confirmation.