MIT’s AI Detects Parkinson’s Early Through Breathing Patterns

MIT's AI department

Abhinav Raj

Abhinav Raj, Writer


MIT’s new artificial intelligence algorithm can detect early onset Parkinson’s disease through a person’s breathing patterns, new research shows.

Pattern recognition enabled by artificial intelligence (AI) is a versatile tool with applications limited only by our imagination.

In 2022, the pattern recognition capability of AI has been powering many life-saving operations—including flood forecasting in inundation-prone areas, timely disaster response and rescue in the event of a natural calamity, and only recently—disease detection and diagnosis.

Researchers at the Massachusetts Institute of Technology (MIT) have developed an artificial intelligence algorithm capable of detecting the early onset of Parkinson’s Disease (PD)—a neurogenerative disease that notoriously evades diagnosis in the early stages.

New research published by MIT in the medical journal Nature Medicine notes the absence of effective biomarkers for the diagnosis of PD—emphasizing the need for an AI model to detect the disease in its preliminary stages and subsequently track its progression.

Built upon a neural network, the AI model monitors a person’s ‘nocturnal breathing signals’ to detect and track the neurological disease. This means that the model observes the breathing patterns of an individual as they sleep overnight, and the neural networks assess the data to determine whether or not they have Parkinson’s.

The methodological underpinnings of the research date back to 1817. MIT electrical engineering and computer science professor Dina Katabiasserts the significance of the findings of Dr James Parkinson in the development of the model.

“This motivated us to consider the potential of detecting the disease from one’s breathing without looking at movements,” explains Katabi.

“Some medical studies have shown that respiratory symptoms manifest years before motor symptoms, meaning that breathing attributes could be promising for risk assessment prior to Parkinson’s diagnosis.”

In a form factor that closely resembles a Wi-Fi router, the home device monitors the respiratory activity of the person (as they sleep) through radio signals. Thus far, the algorithm has been successfully tested on over 7,686 individuals, of which 757 were diagnosed with Parkinson’s.

“In terms of clinical care, the approach can help in the assessment of Parkinson’s patients in traditionally underserved communities, including those who live in rural areas and those with difficulty leaving home due to limited mobility or cognitive impairment,” addsthe senior author of the study.

From detection and diagnosis to progression monitoring, artificial intelligence is assuming an indispensable position in patient management, and by extension, the healthcare infrastructure.