Mount Sinai’s New AI Model Analyzes Entire Nights of Sleep Using 1M+ Hours of Data

Sleep scientists have long sought to decode the mystery of our nightly shuteye, sifting through brain waves, heartbeats, and breathing patterns to understand the various stages we drift through. Now, a new AI model built by researchers at the Icahn School of Medicine at Mount Sinai could shine a light into sleep patterns like never before. Leveraging the same transformer architecture that powers large language models like ChatGPT, this model processes an entire night’s worth of sleep data at once, making it one of the most comprehensive AI tools ever created for sleep analysis. According to a release, 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. Trained on a massive dataset of over a million hours of sleep recordings, PFTSleep could pave the way for faster diagnoses, more accurate sleep staging, and deeper insights into how sleep affects long-term health. Details on the study’s findings were reported in the March 13 online issue of the journal Sleep. “This is a step forward in AI-assisted sleep analysis and interpretation,” says first author Benjamin Fox, a PhD candidate at the Icahn School of Medicine at Mount Sinai in the Artificial Intelligence and Emerging Technologies Training Area. “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.” Traditional sleep studies often depend on human specialists scoring tiny chunks of data, or on existing AI models that can only analyze short snippets at a time. But the new model takes in the whole night in one go. Since it was trained on thousands of full-length sleep recordings (called polysomnograms), it can spot more nuanced patterns that unfold over time and across diverse populations, offering a more consistent and scalable approach to sleep research and potential clinical use, the researchers say. The model was trained using self-supervised learning, a method that lets it extract meaningful patterns from physiological signals, like brain activity or breathing, without needing human labeled data as a guide. “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 PFTSleep is not a replacement for clinical expertise but rather a powerful aid that could accelerate and standardize sleep studies. Future plans include expanding beyond sleep-stage classification into detecting disorders and predicting health outcomes. “This AI-driven approach has the potential to revolutionize sleep research,” said co-senior author corresponding Girish N. Nadkarni, MD, MPH, Chair of the Windreich Department of Artificial Intelligence and Human Health at Mount Sinai. “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 research paper, titled “A foundational transformer leveraging full night, multichannel sleep study data accurately classifies sleep stages,” can be found at this link.

Mar 21, 2025 - 20:09

Mount Sinai’s New AI Model Analyzes Entire Nights of Sleep Using 1M+ Hours of Data

Sleep scientists have long sought to decode the mystery of our nightly shuteye, sifting through brain waves, heartbeats, and breathing patterns to understand the various stages we drift through.

Now, a new AI model built by researchers at the Icahn School of Medicine at Mount Sinai could shine a light into sleep patterns like never before. Leveraging the same transformer architecture that powers large language models like ChatGPT, this model processes an entire night’s worth of sleep data at once, making it one of the most comprehensive AI tools ever created for sleep analysis.

According to a release, 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.

Trained on a massive dataset of over a million hours of sleep recordings, PFTSleep could pave the way for faster diagnoses, more accurate sleep staging, and deeper insights into how sleep affects long-term health. Details on the study’s findings were reported in the March 13 online issue of the journal Sleep.

(Source: Icahn School of Medicine)

“This is a step forward in AI-assisted sleep analysis and interpretation,” says first author Benjamin Fox, a PhD candidate at the Icahn School of Medicine at Mount Sinai in the Artificial Intelligence and Emerging Technologies Training Area. “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.”

Traditional sleep studies often depend on human specialists scoring tiny chunks of data, or on existing AI models that can only analyze short snippets at a time. But the new model takes in the whole night in one go. Since it was trained on thousands of full-length sleep recordings (called polysomnograms), it can spot more nuanced patterns that unfold over time and across diverse populations, offering a more consistent and scalable approach to sleep research and potential clinical use, the researchers say.

The model was trained using self-supervised learning, a method that lets it extract meaningful patterns from physiological signals, like brain activity or breathing, without needing human labeled data as a guide.

“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.”

(Source: Gorodenkoff/Shutterstock)

The researchers emphasize that PFTSleep is not a replacement for clinical expertise but rather a powerful aid that could accelerate and standardize sleep studies. Future plans include expanding beyond sleep-stage classification into detecting disorders and predicting health outcomes.

“This AI-driven approach has the potential to revolutionize sleep research,” said co-senior author corresponding Girish N. Nadkarni, MD, MPH, Chair of the Windreich Department of Artificial Intelligence and Human Health at Mount Sinai. “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 research paper, titled “A foundational transformer leveraging full night, multichannel sleep study data accurately classifies sleep stages,” can be found at this link.