The mistake many make: How to identify early signs of Parkinson's

A new study presents an AI model that identifies early signs of Parkinson's years before physical symptoms, by analyzing medical questionnaires and highlighting symptoms such as loss of smell and sleep disorders.

A new study published in the journal Nature Aging presents an artificial intelligence model called Quest2Dx, developed at Rensselaer Polytechnic Institute (RPI), capable of identifying early signs of Parkinson's disease years before motor symptoms appear. The system analyzes medical questionnaires using large language models to understand the context of each question and answer. The study highlights early symptoms that many tend to ignore, including loss of smell, sleep disorders such as REM sleep behavior disorder, writing difficulties, and mood changes like anxiety. The model identified the 30 most relevant questions from long questionnaires, and the researchers created a shortened questionnaire based on them with nearly identical accuracy. The researchers note that the model still requires improvement and may produce incorrect answers, but they see it as a foundation for developing accessible early diagnostic tools for additional diseases in the future.