New AI tool revolutionizes sleep analysis with comprehensive sleep data

Researchers at the Icahn School of Medicine have developed a powerful AI tool, built on the same transformer architecture used by large language models like ChatGPT, to process an entire night's sleep. To date, it is one of the largest studies, analyzing 1,011,192 hours of sleep. Details on their findings were reported in the March 13 online issue of the journal Sleep [https://doi.org/10.1093/sleep/zsaf061].

The model, called patch foundational transformer for sleep (PFTSleep), analyzes brain waves, muscle activity, heart rate, and breathing patterns to classify sleep stages more effectively than traditional methods, streamlining sleep analysis, reducing variability, and supporting future clinical tools to detect sleep disorders and other health risks.

Current sleep analysis often relies on human experts manually scoring short segments of sleep data or using AI models that are not capable of analyzing a patient's entire night of sleep. This new approach, developed using thousands of sleep recordings, takes a more comprehensive view. By training on full-length sleep data, the model can recognize sleep patterns throughout the night and across different populations and settings, offering a standardized and scalable method for sleep research and clinical use, say the investigators.

This is a step forward in AI-assisted sleep analysis and interpretation. By leveraging AI in this way, we can learn relevant clinical features directly from sleep study signal data and use them for sleep scoring and, in the future, other clinical applications such as detecting sleep apnea or assessing health risks linked to sleep quality."

Benjamin Fox, first author, a PhD candidate at the Icahn School of Medicine at Mount Sinai in the Artificial Intelligence and Emerging Technologies Training Area

The model was built using a large dataset of sleep studies (polysomnograms) that measure key physiological signals, including brain activity, muscle tone, heart rate, and breathing patterns. Unlike traditional AI models, which analyze only short, 30-second segments, this new model considers the entire night of sleep, capturing more detailed and nuanced patterns. Further, the model is trained via a method known as self-supervision, which helps learn relevant clinical features from physiological signals without using human labeled outcomes.

"Our findings suggest that AI could transform how we study and understand sleep," says co-senior corresponding author Ankit Parekh, PhD, Assistant Professor of Medicine (Pulmonary, Critical Care and Sleep Medicine) at the Icahn School of Medicine at Mount Sinai, and Director of the Sleep and Circadian Analysis Group at Mount Sinai. "Our next goal is to refine the technology for clinical applications, such as identifying sleep-related health risks more efficiently."

The researchers emphasize that this AI tool, while promising, would not replace clinical expertise. Instead, it would serve as a powerful aid for sleep specialists, helping to speed up and standardize sleep analysis. Next, the team's research aims to expand its capabilities beyond sleep-stage classification to detecting sleep disorders and predicting health outcomes.

"This AI-driven approach has the potential to revolutionize sleep research," says co-senior corresponding author Girish N. Nadkarni, MD, MPH, Chair of the Windreich Department of Artificial Intelligence and Human Health at the Icahn School of Medicine, Director of the Hasso Plattner Institute for Digital Health, and the Irene and Dr. Arthur M. Fishberg Professor of Medicine. Dr. Nadkarni is also the inaugural Chief of the Division of Data-Driven and Digital Medicine and Co-Director of the Mount Sinai Clinical Intelligence Center. "By analyzing entire nights of sleep with greater consistency, we can uncover deeper insights into sleep health and its connection to overall well-being."

The paper is titled "A foundational transformer leveraging full night, multichannel sleep study data accurately classifies sleep stages."

The study's authors, as listed in the journal, are Benjamin Fox, Joy Jiang, Sajila Wickramaratne, Patricia Kovatch, Mayte Suarez-Farinas, Neomi A. Shah, Ankit Parekh, and Girish N. Nadkarni.