AI Detects Parkinson’s Disease from Nocturnal Breathing

Source: Alexander_Koch/Pixabay

There are over 10 million people worldwide living with Parkinson’s disease (PD) according to the Parkinson’s Foundation. A new study published in Nature Medicine shows how artificial intelligence (AI) machine learning can detect Parkinson’s disease and monitor its progression from nocturnal breathing.

“This work provides evidence that AI can identify people who have PD from their nocturnal breathing and can accurately assess their disease severity and progression,” wrote the researchers from MIT, along with their colleagues at the Mayo Clinic, Massachusetts General Hospital, the Boston University College of Health and Rehabilitation, the University of Rochester Medical Center, and Rutgers University. “Importantly, we were able to validate our findings in an independent external PD cohort. The results show the potential of a new digital biomarker for PD.”

Parkinson’s disease is a progressive neurodegenerative disease where nerve cells in the brain region that controls movement, the basal ganglia, become impaired or die according to the NIH National Institute on Aging. Death and disability due to Parkinson’s disease is increasing at a faster rate globally than any other neurological disorder according to the World Health Organization (WHO).

According to Stanford Medicine, symptoms of Parkinson’s disease may include difficulties in walking or gait, talking (hypophonia or soft voice), swallowing (dysphagia), and unintended involuntary muscle contractions that may cause abnormal postures, twisting, and repetitive movements (dystonia), as well as other symptoms such as depression, anxiety, apathy, mood problems, hallucinations, delusions, psychosis, cognitive problems, sleep issues, fatigue, vision changes, sexual disfunction, constipation, incontinence, low blood pressure, and pain.

For this study, the researchers created a combined dataset of over 11,960 nights with more than 120,000 hours of nocturnal breathing signals from over 750 Parkinson’s disease patients from many sources, including the Michael J. Fox Foundation (MJFF), an MIT observational study, the Mayo Clinic, Massachusetts General Hospital (MGH) sleep lab, the National Institutes of Health (NIH) Udall Center, as well as public sleep datasets from the National Sleep Research Resource.

The datasets were divided into two groups. The first group contained data obtained by using a breathing belt to record nocturnal breathing signals. The second group had datasets collected from a radio device created by MIT to collect nocturnal breathing data without contact. A lower-power radio sensor located in the patient’s bedroom analyzed radio reflections in the environment to obtain the breathing signals.

“By capturing breathing signals using radio signals, our system can run in a completely contactless manner,” the researchers wrote. “We leveraged past work on extracting breathing signals from radio frequency (RF) signals that bounce off people’s bodies.”

The researchers used an AI neural network to predict whether an individual has Parkinson’s disease, and the severity of the disease.

“Our model leverages transfer learning to enable a unified model that works with both a breathing belt and a contactless radio sensor of breathing signals, and transfers the knowledge between different datasets,” the scientists reported.

According to the researchers, their method detects Parkinson’s disease with high accuracy, and the AI biomarker has “shown potential evidence of increased sensitivity to progressive changes in PD.” This development may help reduce the cost and decrease the time required for clinical trials of PD drugs.

“We envision that the system could eventually be deployed in the homes of PD patients and individuals at high risk for PD (for example, those with LRRK2 gene mutation) to passively monitor their status and provide feedback to their provider,” wrote the researchers.

Copyright © 2022 Cami Rosso All rights reserved.