ResMed is bringing artificial intelligence directly into its flagship CPAP machine for sleep apnea. The company’s AirSense 11 device now includes an FDA-cleared setting called Smart Comfort that uses AI to automatically adjust therapy pressure in real time, according to CEO Mick Farrell.

The announcement marks a concrete step in the long‑promised marriage of machine learning and chronic disease management. Sleep apnea, a condition in which breathing repeatedly stops and starts during sleep, affects an estimated 30 million Americans and hundreds of millions worldwide. Continuous positive airway pressure (CPAP) therapy remains the gold standard, but patient adherence has always been a challenge: many users find the fixed pressure settings uncomfortable, especially when they first fall asleep or during transitions between sleep stages.

Smart Comfort is designed to address exactly that friction. Instead of a one‑size‑fits‑all pressure, the AI model analyzes each user’s breathing patterns and adjusts the pressure dynamically. The goal is to make therapy feel more natural so that patients stick with it. Farrell discussed the feature in a recent briefing, emphasizing that the AI setting has already received clearance from the U.S. Food and Drug Administration, meaning it has passed the agency’s scrutiny for safety and effectiveness.

What Smart Comfort actually does

The AirSense 11 is ResMed’s current‑generation CPAP platform, introduced in 2021. It already included features like a touchscreen, built‑in cellular connectivity for remote monitoring, and an earlier comfort algorithm called AutoRamp. Smart Comfort builds on that foundation by adding a dedicated AI layer that can learn a person’s typical sleep profile.

How does the AI work in practice? According to ResMed’s description, the device continuously picks up subtle changes in airflow resistance and breathing effort — data it already collects from its internal pressure sensor and flow generator. The AI algorithm, developed on thousands of anonymized real‑world therapy nights, identifies patterns that indicate whether the user is entering REM sleep, turning over, or experiencing a partial airway collapse. It then tweaks the pressure in small, gradual increments rather than the abrupt jumps that many patients complain about.

Because the adjustment happens automatically, the user doesn’t need to fiddle with settings or even know that the AI is working. The machine simply provides a smoother experience. ResMed says the feature is active by default once enabled, but users can turn it off if they prefer fixed pressure.

Why FDA clearance matters

FDA clearance is not a trivial stamp. For a medical device with an AI component, the agency requires evidence that the algorithm performs safely across a diverse patient population and that changes in its behavior over time are well understood. ResMed’s successful clearance indicates that Smart Comfort met those standards.

This regulatory milestone also signals that the FDA is becoming more comfortable with AI that can adapt therapy without direct clinician intervention — a trend that could accelerate the adoption of similar features in other respiratory devices. The agency has issued guidance for “locked” AI algorithms, which do not change after deployment, as well as for adaptive algorithms that can continue learning. Smart Comfort falls into the adaptive category, albeit with bounds on how much it can deviate from baseline therapy.

The broader push for AI in sleep medicine

ResMed is not the only company exploring AI for sleep apnea. Competitors like Philips and Fisher & Paykel have introduced algorithm‑driven pressure adjustments, and several startups are building cloud‑based analysis tools that mine CPAP data for early signs of treatment failure. But ResMed holds a commanding market share — roughly 40 percent of the global CPAP market — and the integration of AI directly into the device rather than as a separate app or service gives it a clear distribution advantage.

The AirSense 11 already ships with a cellular modem that uploads nightly data to ResMed’s cloud platform, which both patients and clinicians can view through mobile apps and web dashboards. Smart Comfort’s AI logic runs locally on the device, but the company could use the cloud connection to update the algorithm and improve its performance over time.

Farrell has previously spoken about building a “sleep apnea ecosystem” that goes beyond hardware. The company offers a digital coaching program called myAir that provides feedback on mask fit, usage hours, and leak data; Smart Comfort fits into that ecosystem by making the physical therapy itself more adaptive.

What this means for patients

For someone who has struggled with CPAP adherence, Smart Comfort could be the difference between giving up and sticking with treatment. A 2023 meta‑analysis of automatic pressure adjustments found that adaptive algorithms improved average nightly usage by roughly 30 minutes compared with fixed pressure. Even a modest boost in adherence has real clinical meaning: patients who use CPAP at least four hours a night see significant reductions in blood pressure, daytime sleepiness, and cardiovascular risk.

That said, the feature is not a cure‑all. Sleep apnea treatment still depends on correct mask fitting, proper humidity settings, and behavioral factors like avoiding alcohol before bed. AI can’t fix a leaking mask or a patient who simply refuses to wear the device. But by removing one of the most common sources of discomfort — the feeling of being “blasted” with air — the technology may help more people get through the first few weeks, which is the hardest period for new users.

Limitations and open questions

ResMed has not published specific clinical trial data for the Smart Comfort feature in a peer‑reviewed journal, so it’s difficult to independently verify the magnitude of the benefit. The company says the algorithm was trained on “real‑world therapy data,” but the size and diversity of that dataset remain undisclosed. Without that information, it’s not clear how well the AI works for people with unusual breathing patterns or comorbidities such as COPD.

Another concern is that adaptive algorithms might mask or delay the detection of disease progression. If the machine constantly compensates for worsening sleep apnea by raising pressure, a clinician reviewing cloud data could miss signs that a patient’s condition is becoming more severe. ResMed’s platform does alert clinicians to major changes in the apnea‑hypopnea index (AHI), but the threshold for what counts as a “major” change is set by the provider, not the algorithm.

Finally, there is the question of cost. The AirSense 11 is already priced at a premium, and Smart Comfort is included standard — there is no extra fee for the AI feature. But insurance reimbursement for CPAP equipment typically covers the machine itself, not software upgrades. If ResMed later decides to gate advanced AI features behind a subscription model, patients could face new out‑of‑pocket costs.

What comes next

Farrell’s discussion of Smart Comfort signals that ResMed sees AI as a core differentiator for its next generation of products. The company is also exploring AI for diagnosis — using home sleep test data to automatically classify severity — and for predicting which patients are at risk of abandoning therapy. Those tools are likely still in development, but the FDA clearance of Smart Comfort gives ResMed a beachhead.

For now, anyone who buys a new AirSense 11 will get the AI feature out of the box. Existing AirSense 11 users can enable Smart Comfort through the device’s settings menu after a firmware update. The company has not announced plans to bring the feature to older models like the AirSense 10.

Sleep apnea is a chronic condition that is often underdiagnosed and undertreated. Tools that reduce the daily burden of therapy — without adding complexity for the patient — have the potential to improve population health in a meaningful way. ResMed’s Smart Comfort is a small, specific application of AI, but it represents the kind of incremental improvement that, multiplied across millions of users, adds up to a real difference.