A new dataset could help researchers develop AI tools to detect Parkinson’s remotely using motor and memory data collected through home computers.
Parkinson’s can cause motor symptoms including slowness and tremor, as well as non-motor symptoms such as memory problems and digestive complaints.
Clinical assessment by an expert is currently the gold standard for diagnosing Parkinson’s and monitoring disease severity.
Researchers said in-person assessments can create logistical difficulties, particularly for people who do not live near specialist centres. Reliance on a human evaluator also makes the assessments inherently subjective.
As computers and smart devices have become more common in homes, researchers have been exploring whether they could be used to monitor Parkinson’s remotely, offering greater convenience and objectivity.
Artificial intelligence uses algorithms to identify patterns in large amounts of data and requires large, well-defined datasets for training and testing.
Researchers used an online platform to generate a new dataset for training AI in Parkinson’s assessments.
Participants completed movement-based assessments using a computer mouse and keyboard, followed by a memory test. The tasks could be completed by anyone with a computer and internet access.
Researchers extracted dozens of measures from the tasks, including the precise timing of keystrokes and variations in mouse movements.
The dataset, called RobustPDx, included data from 261 participants: 73 with self-reported Parkinson’s, 33 with self-reported suspected Parkinson’s and 155 controls without the condition.
They also recorded participants’ demographic information, handedness and the type of computer they were using.
“As remote digital health assessments increasingly rely on AI, it is important to evaluate whether these models remain robust across real-world sources of variation, including differences in device type and handedness,” the researchers noted.
The team hopes RobustPDx will support future studies developing AI tools for remote diagnosis and monitoring of Parkinson’s.
Researchers said the dataset could also be used to examine the robustness and reliability of AI models across different users, devices and patterns of interaction.
