Developed by Diego Guarin, PhD, an assistant professor of applied physiology and kinesiology at the University of Florida (UF), the system applies machine learning to analyse video recordings of patients performing the finger-tapping test, a standard test for Parkinson’s disease that involves quickly tapping the thumb and index finger 10 times.
“By studying these videos, we could detect even the smallest alterations in hand movements that are characteristic of Parkinson’s disease but might be difficult for clinicians to visually identify,” Guarin said in a statement. “The beauty of this technology is that a patient can record themselves performing the test, and the software analyses it and informs the clinician how the patient is moving so the clinician can make decisions.”
Parkinson’s disease is a brain disorder that affects movement and can result in slowness of movement, tremors, stiffness, plus difficulty with balance and coordination. Symptoms usually begin gradually and worsen over time. According to UF, there is no specific lab or imaging test that can diagnose Parkinson’s disease, but a series of exercises and manoeuvres performed by the patient helps clinicians identify and evaluate the severity of the disorder.
MORE FROM ARTIFICIAL INTELLIGENCE
The rating scale most used to follow the course of Parkinson’s disease is the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale. Guarin said that despite its reliability, the rating is restricted to a five-point scale, limiting its ability to track subtle changes in progression and is prone to subjective interpretations.
“We found that we can observe the same features that the clinicians are trying to see by using a camera and a computer,” said Guarin. “With help from AI, the same examination is made easier and less time-consuming for everyone involved.”
Guarin said the automated system has also revealed previously unnoticed details about movement using precise data collected by the camera, like how quickly the patient opens or closes the finger during movement and how much the movement properties change during every tap.
“We’ve seen that, with Parkinson’s disease, the opening movement is delayed, compared to the same movement in individuals that are healthy,” said Guarin. “This is new information that is almost impossible to measure without the video and computer, telling us the technology can help to better characterise how Parkinson’s disease affects movement and provide new markers to help evaluate the effectiveness of therapies.”
To perfect the system, the team tapped into UF’s HiPerGator supercomputer to train some of its models.
“HiPerGator enabled us to develop a machine learning model that simplifies the video data into a movement score,” said Guarin. “We used HiPerGator to train, test, and refine different models with large amounts of video data, and now those models can run on a smartphone.”
In addition to placing this technology in the hands of neurologists and other care providers, Guarin is working with UFIT to develop it into an app for mobile devices, allowing individuals to assess their disease over time at home.
The team’s findings are detailed in IEEE Transactions on Neural Networks.
Comment: Bristol can provide blueprint for heat network success
I was thinking that the outlet temperature and energy from a (thermal) power station would depend upon its source temperature. And so an SMR (usually...