Connecting Sleep Data to Clinical Practice

💤 COMISA: The hidden barrier ⟶ to Longevity Progress ❓What if your most committed patients, tracking nutrition, exercising consistently, practising mindfulness, or peri/menopausal women...still feel stuck? Energy dips. Glucose variability. Brain fog. The missing link? Often... sleep quality. In particular: COMISA. The co-occurrence of insomnia and obstructive sleep apnoea (OSA). 💫 Why this matters in longevity & lifestyle medicine In my clinical practice, I’ve seen patients doing everything right. Yet progress plateaus. Labs look fine, habits are solid… but fatigue and metabolic instability persist. When we dug deeper, the culprit was clear: COMISA. Once both components, insomnia and airway management, were addressed, progress accelerated. Energy stabilised.  Cravings reduced.  Glycaemic variability improved. The turning point was... Restorative sleep architecture. 📊 COMISA affects approximately 30–50% of people with OSA and 30–40% of those with insomnia. (Sweetman et al., Sleep Medicine Clinics, 2022) 🌍 In large population-based studies, COMISA prevalence varies: 11.4% in Switzerland, 9.1% in India, and 1.7% in Benin — showing global variation. (Zhou et al., Sleep Health, 2024) 🧠 COMISA is associated with greater impairment in sleep quality, mood, and daytime functioning than insomnia or OSA alone. (Garbarino et al., Frontiers in Neurology, 2023) ❤️ Patients with COMISA show poorer CPAP adherence and higher cardiometabolic risk compared to those with OSA alone. (Krakow et al., Journal of Clinical Sleep Medicine, 2021) When sleep is disrupted on multiple fronts, even the most evidence based lifestyle plans...nutrition, movement, stress management... can underperform. 🔎 Screening & integration 🇬🇧 In the UK, COMISA can be screened by pairing insomnia and OSA risk tools: 🛏 Insomnia: Insomnia Severity Index (ISI) + 💨 OSA: STOP-Bang Questionnaire (snoring, tiredness, observed apnoea, high blood pressure, BMI, age, neck circumference, gender) ⟶ If risk is flagged, primary care (or other services e.g. Tier 3 Weight Management Services) can refer for home sleep studies or sleep clinic evaluation. ⟶ NHS pathways recommend combining sleep study data with insomnia assessment to capture both sides. (NHS Sleep Medicine Pathway BNSSG ICB, 2024) Don’t treat insomnia or OSA in isolation. When both are managed together, outcomes improve across metabolic, cognitive, and emotional domains. 🔎 So what becomes possible when we address COMISA? ⟶ Better adherence to lifestyle plans (energy + cognitive clarity return) ⟶ Improved metabolic + inflammatory profiles ⟶ Enhanced emotional resilience + motivation ⟶ A stronger foundation for healthy ageing ❓ For my clinician peers: - Are you routinely screening for COMISA in your patients? - What barriers or successes have you seen? #LongevityMedicine #LifestyleMedicine #SleepHealth #COMISA #SleepApnoea #Insomnia #Healthspan #IntegrativeCare

𝗖𝗶𝗿𝗰𝗮𝗱𝗶𝗮𝗻 𝗵𝗲𝗮𝗹𝘁𝗵 𝗶𝘀 𝗰𝗮𝗿𝗱𝗶𝗼𝗺𝗲𝘁𝗮𝗯𝗼𝗹𝗶𝗰 𝗵𝗲𝗮𝗹𝘁𝗵. The American Heart Association’s new scientific statement makes it plain: it is not only how much we sleep, but also when we sleep, when we see light, and when we eat. Disrupted body clocks are linked to obesity, type 2 diabetes, hypertension, and cardiovascular disease. 𝗪𝗵𝗮𝘁 𝘁𝗵𝗲 𝗲𝘃𝗶𝗱𝗲𝗻𝗰𝗲 𝘀𝘁𝗿𝗼𝗻𝗴𝗹𝘆 𝘀𝘂𝗴𝗴𝗲𝘀𝘁𝘀 - Irregular sleep timing and variable schedules correlate with higher cardiometabolic risk, independent of total sleep time. - Shift work and “social jet lag” disrupt endocrine and autonomic rhythms that regulate glucose, blood pressure, and lipids. - Aligning daily behaviors to the internal clock is a promising prevention strategy, although more causal trials are needed. 𝗪𝗵𝗮𝘁 𝘁𝗵𝗶𝘀 𝗺𝗲𝗮𝗻𝘀 𝗶𝗻 𝗰𝗹𝗶𝗻𝗶𝗰𝗮𝗹 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗲 𝗮𝗻𝗱 𝗽𝘂𝗯𝗹𝗶𝗰 𝗵𝗲𝗮𝗹𝘁𝗵 - Stop asking only “how many hours.” Start asking “when do you sleep, when do you eat, when is your light exposure.” - Treat sleep as multidimensional: duration, timing, regularity, quality, and daytime alertness all matter. - Address inequities that drive misalignment: rotating shifts, late-evening service work, light pollution, and neighborhood safety that limits morning light and evening exercise. 𝗟𝗼𝘄-𝗹𝗶𝗳𝘁 𝗶𝗻𝘁𝗲𝗿𝘃𝗲𝗻𝘁𝗶𝗼𝗻𝘀 𝘆𝗼𝘂 𝗰𝗮𝗻 𝘀𝘁𝗮𝗿𝘁 𝘁𝗵𝗶𝘀 𝘄𝗲𝗲𝗸 - Morning outdoor light within an hour of waking, even on cloudy days. - Fixed sleep and wake targets, 7 days a week - Daytime-weighted eating window, avoid heavy late meals, cut bright light at night. - Exercise most days, place vigorous sessions earlier when possible. These steps support circadian alignment, which the AHA highlights as a preventive pathway for cardiometabolic risk reduction. 𝗙𝗼𝗿 𝗺𝘆 𝗰𝗼𝗹𝗹𝗲𝗮𝗴𝘂𝗲𝘀 𝗶𝗻 𝗰𝗮𝗿𝗱𝗶𝗼𝗹𝗼𝗴𝘆, 𝗲𝗻𝗱𝗼𝗰𝗿𝗶𝗻𝗼𝗹𝗼𝗴𝘆, 𝗽𝗿𝗶𝗺𝗮𝗿𝘆 𝗰𝗮𝗿𝗲, 𝗮𝗻𝗱 𝘀𝗹𝗲𝗲𝗽: - Integrate circadian screening in vitals, history, and lifestyle counseling. - Embed “clock-aware” recommendations into discharge instructions and chronic disease pathways. - Prioritize research and QI projects that test timing-based interventions alongside standard therapy. The statement’s takeaway is clear: circadian-aware care is a tractable lever for prevention, with compelling observational data and growing mechanistic support. 𝗖𝗮𝗹𝗹 𝘁𝗼 𝗮𝗰𝘁𝗶𝗼𝗻 If your team treats or counsels on blood pressure, lipids, glucose, weight, or sleep apnea, make circadian health part of the protocol. Share this with your care coordinators, employer health programs, and shift-work communities. Patients cannot out-medicate a misaligned clock.

Today in Nature Medicine we report that AI can predict 130 diseases from 1 night of sleep 🛌. We trained a foundation model (#SleepFM) on 585K hours of sleep recordings from 65K people—brain, heart, muscle & breathing signals combined. AI learns the language of sleep! Paper: https://lnkd.in/grpRD3Qp Open source code: https://lnkd.in/g_MqFCm6 Participants are linked to their EHR. SleepFM predicts risks for diverse diseases--including dementia, heart failure, kidney disease, and stroke--years before clinical diagnosis. It substantially outperforms using demographic features, which are strong predictors. SleepFM uses a new architecture to integrate multimodal sleep time-series data. CNNs learn local features, transformers aggregate information across time + channels, and leave-one-modality-out contrastive learning trains robust representations. This design generalizes across sites and diverse populations. We spend 1/3 of our lives sleeping but it has been underexplored with AI. Most work focuses on narrow tasks like sleep staging and apnea detection. By learning a holistic representation of sleep, SleepFM opens new doors for studying the science and medicine of sleep. Truly wonderful collaboration with Emmanuel JM Mignot's lab, led by Rahul Thapa and Magnus Ruud Kjaer! Thanks to all the awesome collaborators: Bryan He, Ian Covert, Hyatt Moore, Umaer Hanif, Gauri G., M Brandon Westover, Poul Jennum, Andreas Brink-Kjær 👏

Stanford AI predicts dementia from one night of sleep data. 85% accuracy. No brain scan. No blood test. Just a sleep study. This changes everything about early detection. The study (published January 9, 2026): Analyzed sleep data from sleep studyies. Measured brain-heart-breathing coordination during sleep Detected subtle desynchronization invisible to humans Predicted dementia diagnosis 3+ years before symptoms 85% accuracy from one night. What the AI detected: Your brain, heart, and breathing normally synchronize during sleep. In early Alzheimer's, this synchronization breaks down. Years before memory symptoms. The changes are tiny. You can't feel them. Standard sleep studies miss them. AI found the pattern. Why this matters: Current dementia screening requires: ↳ Symptoms already present ↳ Cognitive testing showing impairment ↳ Expensive biomarker testing or scans This approach finds risk before symptoms. What it could enable: Screen everyone over 50 during annual checkup ↳ Download sleep data from watch ↳ Run through AI algorithm ↳ Flag high-risk individuals for further testing Catch Alzheimer's in preclinical stage ↳ When prevention works best ↳ Before irreversible damage ↳ 10-15 year intervention window The limitations: Still in research phase Needs validation in larger populations Not FDA-approved yet Requires specific sleep metrics from devices But the proof of concept is there. My prediction: Within 5 years, your annual physical will include: "Bring your sleep data from your watch." AI screens it overnight. High-risk patients get blood biomarker testing. Positive blood test gets preventive intervention. We find Alzheimer's 15 years before symptoms. We prevent it from progressing. Mass screening becomes possible. This is how we'll finally get ahead of dementia. Not by treating it at 75. By catching it at 60. ⁉️ Would you want your sleep data screened for dementia risk? ♻️ Repost if you believe in catching disease before symptoms 👉 Follow me (Reza Hosseini Ghomi, MD, MSE) for emerging diagnostics Citations: Sommerlade L. Multi-modal Sleep Analysis for Early Dementia Detection Using Machine Learning. Nature Medicine. 2026.

Google just turned Fitbit into an AI health platform. Your wearable now wants your medical records. At The Check Up, Google's annual health event, Fitbit announced the biggest update to its Personal Health Coach since launch. Medical records integration, advanced sleep science, metabolic research, and continuous glucose monitoring, all feeding into a Gemini-powered AI coach. From a fitness tracker to a personal health operating system on your wrist. Here is what is new: → Starting next month, U.S. Public Preview users can link their medical records to the Fitbit app, lab results, medications, visit history, all in one place → Identity verification through CLEAR and IAL2-certified standards, a selfie and a valid ID to automatically locate and sync records across providers → The AI coach can now contextualize your health questions using your actual medical history, not generic answers, but personalized guidance based on YOUR data → Continuous glucose monitor (CGM) integration via Health Connect, ask the coach how a workout or a slice of pizza impacts your glucose levels → 15% improvement in sleep staging accuracy, trained on diverse clinical datasets aligned with gold-standard measurements → A reimagined Sleep Score that evaluates specific aspects of rest, not just duration, but how long it took you to fall asleep and where to focus for better recovery → A new "Get care now" Fitbit Lab study in partnership with Included Health to assess how conversational AI can help navigate virtual care visits The research backing is serious. Google's pioneering study on predicting insulin resistance using wearable data was just published in Nature. The model combines wearable signals with routine blood biomarkers to predict insulin resistance with 80% auROC, and 93% sensitivity in the most at-risk populations. This is foundational work for early type 2 diabetes screening at scale. The privacy architecture: → Medical records securely stored with Fitbit, under your control → You decide how data is used, shared, or deleted → Medical records are not used for ads → Secure sharing with family or providers via Smart Health Link URL or QR code Now zoom out. Google is not just adding features. Google is building the full stack: → Wearable hardware (Fitbit + Pixel Watch) → Real-time biometric data (heart rate, sleep, activity, SpO2) → Clinical validation (Nature-published metabolic research) → Medical records integration (lab results, medications, visit history) → CGM data (continuous glucose monitoring via Health Connect) → AI reasoning layer (Gemini-powered Personal Health Coach) → Care navigation (Included Health partnership for virtual visits) From sensor to insight to action to care, in one ecosystem. Now Google turns its 30+ million Fitbit users into a health data platform. The difference: Google already has the wearable on your wrist. The research is already published. The clinical foundation is already built. Source & picture: google

The patient was waking up every 2-3 hours, never felt rested, and was exhausted all the time. So you ordered the sleep study because you were pretty sure this was sleep apnea. Then the report came back saying: “No obstructive sleep apnea”. And now you're sitting with the patient who’s still struggling, a normal-ish sleep study, and not much of a roadmap for what to do next. This scenario comes up a lot in clinical practice, and it's exactly what inspired me to make this week's video. One of my newsletter readers had written to me asking: "Do you have any information on treatment for interrupted sleep patterns, for example, waking every 2-3 hours in the context of a negative home sleep study?" So this week, I’m sharing what to do next, including the 4 broad categories that drive most sleep fragmentation, a few clinical case examples, and a personal sleep fragmentation story of my own. Fair warning: I recorded this while experiencing an upper respiratory infection, so please excuse the congestion. It actually felt timely, because nasal congestion is one of those commonly overlooked factors that can fragment sleep, and I got a firsthand reminder of that while making this video! I’ll put the video link in the comments. —— Hi! I'm Nishi Bhopal MD 🇨🇦🇺🇸Canadian expat. Dog lover. Psychiatrist & Sleep Doc. ✅ I teach physicians & therapists about clinical sleep medicine 🩺 Founder of a private practice called Pacific Integrative Psychiatry ✓ Information only, not medical advice #physicians  #therapists #insomniatreatment #sleepmedicine  #sleepmedicinedoctor #clinicalsleep #clinicalmedicine #practiceofmedicine  #privatepractice  #medicaleducation #CME

Digital sleep clinic: assessing efficacy of continuous positive airway pressure through sleep staging via connected devices: a study protocol Link: https://lnkd.in/erp263NZ Background Obstructive sleep apnea (OSA) is a multisystemic chronic disease with disabling symptoms, cardiometabolic comorbidities and reduction in physical activity. Continuous positive airway pressure (CPAP) is the standard treatment for OSA. Only a few studies have characterized trajectories of sleep parameters upon initiation of CPAP and these are limited to one or two nights of polysomnographic recording in a sleep laboratory. This is due to the cost of carrying out these studies and poor tolerance by patients of multiple nights of polysomnographic recordings. No study has characterized sleep over multiple nights before and after CPAP initiation, assessing the multidimensional efficacy of CPAP on patient reported outcomes, objective and subjective sleep quality, oximetry, glucose control and physical activity. New digital technologies enable overnight sleep studies over several nights in the patient’s home, with a reliability of sleep characterization equivalent to polysomnographic recording. The primary aim of this study is to investigate objective slow wave sleep (SWS or N3) quality before CPAP and during the first month of the treatment. Secondary objectives are to assess changes in the following parameters before CPAP and during the first month of the treatment: other objective sleep parameters and sleep stages evolution (W, N1, N2 and REM), nocturnal oxygen desaturations, 24-h blood glucose profile, daily physical activity (the daily steps count), and patient reported outcomes. Methods Seventy patients prescribed CPAP for OSA will be recruited at Grenoble Alpes University Hospital (France) and monitored for 5 weeks using validated innovative wearable connected devices (the Dreem Health | Sunrise 3 headband, a pedometer, an oximeter, and a continuous glucose sensor) enabling them to track their own sleep and physiological parameters at home before and after CPAP initiation. Discussion By pooling data from the CPAP telemonitoring and other connected devices we should be able to follow the multidimensional trajectories of patients after the initiation of CPAP. This will enable us to determine whether objective changes in sleep parameters in the first few weeks of CPAP treatment are associated with improvements in daytime sleepiness, quality of life, treatment adherence, glucose control and physical activity. The data will provide integrated markers of treatment efficacy and will allow adapted personalized management of OSA in the short and long-term. #sleep #sleep2025 #sleepapnea #osa #cpap #sleeptrends #health #healthcare #sleepresearch #sleeptech #hme

Some of the most important signals in health may come from a place we’ve traditionally overlooked: during sleep. A new study, published by Michael Chee and his team in PLOS Digital Health, looked at whether nighttime photoplethysmography from Oura Ring could be used to estimate vascular aging, a meaningful marker of cardiovascular risk. The researchers found that Oura Ring performed comparably to a clinical-grade fingertip sensor in estimating vascular age from pulse waveforms collected overnight. One concrete detail that stood out to me: the deep learning model achieved a mean absolute error of about 7.25 years using ring data, versus 6.28 years with the clinical-grade device, with no statistically significant difference between them. This is a significant proof point that scalable, longitudinal, low-friction sensing may help us understand cardiovascular health earlier, more continuously, and in the context of daily life. If we want to move healthcare upstream, we need tools that meet people where they are — and sometimes that means learning from the physiology we can observe at night. To me, that’s where digital health becomes genuinely useful: not louder, but earlier, steadier, and more actionable. Read more on the Pulse blog: https://lnkd.in/eRyNd_qT Full paper: https://lnkd.in/eT_DgE5P

𝗬𝗼𝘂𝗿 𝗕𝗼𝗱𝘆 𝗦𝗵𝗼𝘄𝘀 𝗜𝘁𝘀 𝗙𝘂𝘁𝘂𝗿𝗲 𝗪𝗵𝗶𝗹𝗲 𝗬𝗼𝘂’𝗿𝗲 𝗔𝘀𝗹𝗲𝗲𝗽   𝘠𝘰𝘶𝘳 𝘣𝘰𝘥𝘺 𝘬𝘯𝘰𝘸𝘴 𝘸𝘩𝘢𝘵’𝘴 𝘤𝘰𝘮𝘪𝘯𝘨 𝘭𝘰𝘯𝘨 𝘣𝘦𝘧𝘰𝘳𝘦 𝘺𝘰𝘶 𝘥𝘰. 𝘔𝘦𝘥𝘪𝘤𝘪𝘯𝘦 𝘶𝘴𝘶𝘢𝘭𝘭𝘺 𝘧𝘪𝘯𝘥𝘴 𝘰𝘶𝘵 𝘵𝘰𝘰 𝘭𝘢𝘵𝘦.   In early 2026, Stanford Medicine researchers unveiled SleepFM—an AI model trained on nearly 600,000 hours of polysomnography data from 65,000 people. By reading a single night of sleep—brain waves, heart rhythm, breathing, and micro-movements—the model detects biological patterns associated with elevated long-term risk across 100+ diseases, including heart disease, dementia, cancer, and stroke.   This exposes a structural flaw in modern healthcare:  • We wait for symptoms instead of signals.   𝗦𝗹𝗲𝗲𝗽𝗙𝗠 𝗽𝗼𝗶𝗻𝘁𝘀 𝘁𝗼 𝗮 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗿𝗲𝗮𝗹𝗶𝘁𝘆:  • Disease progression leaves measurable fingerprints years before diagnosis  • Sleep captures system-level physiology no clinic visit ever sees  • One night of high-resolution data can be more informative than years of sporadic checkups  • AI’s real power is not automation, but early truth extraction  • This is not about better sleep scores or optimization culture.  • It’s about shifting healthcare from reaction to anticipation.   The future of medicine won’t start in hospitals. It will start in places where the body stops performing and starts revealing. Source & details: Stanford Medicine SleepFM study (Nature Medicine, 2026): https://lnkd.in/gxBYePw5 https://lnkd.in/gnYB53sJ https://lnkd.in/gurVfR3c #ArtificialIntelligence #MultimodalAI #FoundationModels #BrainyNeurals #AIinHealthcare #ClinicalAI #ComputerVision #EdgeAI #MultimodalAI #SleepScience #PreventiveMedicine #EarlyDetection #Neuroscience #FutureOfMedicine

Your sleep tracker is measuring the wrong things. Eight hours. Sleep score 85. Deep sleep percentage. None of these metrics predicted a single disease in the largest sleep study ever conducted. A paper dropped this week in Nature Medicine. 585,000 hours of sleep data. 65,000 participants. 130 diseases predicted from one night. Parkinson's disease: 93% accuracy. Dementia: 85%. All-cause mortality: 84%. From polysomnography. Not from your wrist. The signals that matter are not duration. They are the interactions between brain waves, heart rhythm, respiratory patterns, and muscle activity. Simultaneously. Across the night. Your consumer wearable captures almost none of this. We have spent a decade optimizing for metrics that correlate with nothing. Meanwhile, the predictive power was in data we already collect in sleep clinics. We just never analyzed it properly. Foundation models changed the game. Not because AI is magic. Because 585,000 hours of unlabeled data taught the model what disease looks like in sleep physiology. Before symptoms appear. Before diagnosis. Before it is too late to intervene. The infrastructure exists. Sleep labs run thousands of studies per year. The question is no longer whether sleep predicts disease. The question is why we are not using it. Full analysis in the article below. The implications for digital biomarkers, drug development, and clinical practice are significant.