Icahn School of Medicine at Mount Sinai’s Post

🌙 We're excited to share groundbreaking results from our latest research published in the journal Sleep. Our team at Mount Sinai has developed PFTSleep, a foundational transformer AI model trained on full-night, multichannel sleep studies. Remarkably, this model achieves state-of-the-art accuracy in classifying sleep stages across diverse datasets—representing over 1 million hours of polysomnography data! 🛌 Why is this significant? Sleep disorders impact nearly 30% of adults in the U.S., yet only 5% receive accurate diagnoses. Traditional sleep stage classification is time-intensive and subject to human variability. Our AI-driven approach, using representations learned independently from sleep stage classification, offers both precision and scalability. 🤖 The PFTSleep model leverages deep-learning techniques similar to those behind tools like ChatGPT, but tailored specifically for physiological signals collected during sleep. It not only matches expert human scoring but also demonstrates adaptability across multiple cohorts, highlighting its potential for widespread clinical adoption. 📈 We envision that these robust, AI-derived sleep representations will become foundational—not just for accurate sleep stage classification, but also for predicting other important clinical outcomes related to sleep health. 🔗 Curious to explore the full paper? Dive deeper here: https://bit.ly/3ZQw5Y9 https://bit.ly/3ZQw5Y9 💡 We're keen to engage with sleep researchers, clinicians, and AI enthusiasts—your thoughts and insights would be invaluable! Hasso Plattner Institute for Digital Health at Mount Sinai Hasso Plattner InstituteHasso Plattner Foundation Icahn School of Medicine at Mount SinaiMount Sinai Health SystemAI and Human Health at Mount Sinai Benjamin F.Joy Jiang Patricia Kovatch Mayte Suarez-FarinasNeomi Shah Ankit Parekh #AI #SleepResearch #Transformers #DigitalHealth #MountSinai