New positive airway pressure algorithms show increased comfort and reduced side effects. The unanswered question is whether they will meaningfully change long-term adherence.
By Sree Roy
For anyone who has been in sleep medicine long enough, the buzz surrounding several newly announced CPAP algorithms may be reminiscent of the promises of auto-CPAP (APAP) decades ago. Personalized therapy! Lower pressures! Fewer side effects! Reduced need for in-person visits! However, two decades later, the evidence that APAP is better than CPAP is, at best, mixed.1 Adherence also remains much lower than desired.
Once more, the field is seeing an influx of new and emerging technologies that strive, at their heart, to solve the perennial problem: long-term adherence. But, with limited rollouts so far, data is unavailable to determine whether that outcome will be achieved. Early indicators, however, including increased patient comfort, higher usage compared to legacy algorithms (at least in the short term), and fewer annoyances like mask leaks, hold the promise of a new generation.
“I’m really excited by what the future of personalization can bring to sleep health,” says Charles Hartson, vice president of product management at Resmed. “It seems like we’re at a real inflection point of change in terms of those outcomes and what those solutions can bring. And I’m really excited to see the industry move forward over the next few years.”
Algorithms that could move the industry forward include the newly launched Resmed Smart Comfort and React Health Smart-A (short for Smart-Auto), the soon-to-be-available SleepRes KairosPAP (KPAP), and the investigational NovaResp cMAP. For patients who are newly diagnosed with obstructive sleep apnea this year, as well as those due for a new device, sparkly new options proliferate.
Personalization Powered by Artificial Intelligence
Advances in artificial intelligence (AI) and machine learning help explain why multiple medtech innovators debuted changes simultaneously.
Food and Drug Administration (FDA)-cleared Smart Comfort, Hartson explains, “is a big milestone for us,” enabling Resmed to step forward in personalization. It leverages millions of nights of data to suggest a starting point for therapy based on patient inputs, including their age, apnea-hypopnea index (AHI), and sex.
Before the availability of Smart Comfort, identifying the optimal settings for an individual could be time-consuming. “They are available one by one, but they are stuck behind various user interfaces,” Hartson says. Smart Comfort quickly predicts these non-prescription settings to provide “the best first night experience possible for the patient,” according to Hartson.
“Today, a lot of solutions are very population-based—one size fits all,” he says. “What this allows us to do is to really start to deliver more personalized solutions and experiences for patients to make that journey to start on CPAP therapy the best one feasible.”
Retrospective comparisons of Smart Comfort-recommended settings to default configurations showed improvements in patient therapy engagement, including increased nightly usage and additional usage days, while maintaining comparable residual AHI and mask leak, according to data Resmed provided to the FDA.2
Also AI-enabled, NovaResp cMAP is designed to anticipate sleep-breathing events and even to prevent them. The software demonstrated a 15% improvement in adherence in a blinded 100-patient trial comparing cMAP to standard APAP—a gap that widened after three months as standard support from the durable equipment manufacturer decreased. The algorithm also reduced the time sleep techs spent coaching patients by 50%, as users logged more sleep time and requested fewer mask changes.3 “The results of the trial have been very promising,” says Hamed Hanafi, PhD, NovaResp founder and CEO.
NovaResp continues to refine its approach by researching a transfer-learning approach (a type of machine learning workflow) to personalize the algorithm to individual patients using a limited amount of their PAP data. One study found that personalized cMAP significantly improved the models’ ability to identify sleep-breathing events, a 38% increase, on average.4
New Avenues of Pressure Relief
Lowering the centimeters of water pressure is a target of many of these algorithms, though each approaches it differently.
William Noah, MD, founder and chief science officer of SleepRes Inc and inventor of the KPAP algorithm, identified a “Kairos”—a Greek term for a precise or critical time—which he defines as the last second of the respiratory cycle.
“CPAP is going to give you that pressure the whole four seconds,” Noah says. “But what I discovered from a physiological point of view is that there’s this critical time, which is the last second. If you gave the full therapy pressure just there, it would be equal to the pressure over the whole four seconds. The other three seconds are unnecessary and lead to side effects and discomfort.” A peer-reviewed clinical trial backs that up, finding that awake-users considered KPAP to be subjectively more comfortable than CPAP. In the same trial, KPAP was found to have similar efficacy in lowering AHI and also reduced unintentional leak.5
NovaResp’s cMAP algorithm focuses on lowering the pressure of therapy (by about 20%) by identifying the minimum pressure required to prevent an event before it occurs. By predicting upcoming apneas, the system can perform gentle pressure maneuvers, avoiding the high-pressure “chasing” of events typical of standard APAP.
In its new G3X foam-free device, React Health inserted a new algorithm that can apply “micro-adjustments” rather than sudden pressure spikes, according to Kim Applegarth, RRT, director of the React Health Institute. To prevent arousals, pressure increases are often divided across several breaths, with smaller increments used as the overall pressure builds. “We know clinically that outcomes are better if we can treat at those lower pressures and also with fewer pressure swings,” Applegarth says.
Clinicians can further personalize therapy on the G3X using “fast,” “standard,” or “soft” sensitivity settings for patients with specific needs, such as a high AHI or a high sensitivity to pressure changes.
Notably, none of the manufacturers claim these algorithms are more effective at treating sleep apnea than standard CPAP. Instead, they provide evidence for equal efficacy while significantly improving the patient experience. If increased comfort leads to long-term adherence, however, that could open the door to better outcomes, for example, improved sleep architecture—for which NovaResp has seen preliminary evidence via SleepImage’s Sleep Quality Index.
Prescribing New Devices
For newly diagnosed patients, the appeal of these algorithms is clear: easing therapy onboarding. Resmed, for instance, is focusing the initial launch of its Smart Comfort on new patients using the AirSense 11 and the myAir app. While the settings are available to anyone on the platform, the AI-driven recommendations are explicitly intended to optimize the crucial “first-night experience,” Hartson notes. “I believe in a future of healthcare that’s really personalized,” he says, adding that all patients should have the opportunity to find the best combination of parameters for them.
But what about the millions of patients already using legacy devices? For sleep clinicians, the decision to transition an existing patient to a new algorithm requires careful consideration. Applegarth takes a pragmatic approach. “If a patient is doing well, the data looks well, and the patient is not complaining, that says to me we’re treating the patient,” she says. “But if a patient is struggling with compliance, or not getting the results they used to get as their condition changes, that’s a patient that should consider trying one of these newer devices.”
Other developers argue that the benefits of this new generation extend even to long-term, adherent users, particularly when they come due for a new device (about once every five years). Noah points out that a device equipped with KPAP “still does everything theirs does, but it also has KPAP for those who want it.” He suggests that the algorithm’s potential to prevent aerophagia and treatment-emergent central sleep apnea justifies its use even for patients who currently tolerate standard pressure.6 In a three-month usability trial consisting entirely of long-term, adherent users, Noah notes that a frequent comment when patients had to return the KPAP machines was, “Hey, I want that back.”
NovaResp saw similar enthusiasm in a 50-patient comfort study consisting of long-term CPAP users. These users rated the cMAP therapy higher, showed improved objective sleep quality metrics, and reported anecdotal benefits like bed partners noticing a quieter mask. “I think there’s a huge potential there for even long-term patients to experience better therapy,” Hanafi says. (NovaResp separately tracked 100 treatment-naive patients in an adherence trial to prove its impact on new users.)
Still, predictive and pressure-variable algorithms raise clinical questions about reliability. If the inputs are incorrect—whether from patient, physician, sleep tech, or algorithm error—the delivered pressure could be inadequate. CPAP, with its unrelenting pressure, is a sure thing.
Synergy and the Future of Therapy
An intriguing possibility is for these algorithms to work in tandem. As one example, because KPAP changes the pressure modulation within a breath, and NovaResp or React Health’s algorithms change the baseline therapy pressure, there is a theoretical synergy.
As the field moves toward precision sleep medicine, the focus is shifting toward data-driven learning that adapts to the patient’s evolving condition, including changes in weight or medication. Whether these new algorithms will finally close the adherence gap remains to be seen, but sleep clinicians should stay current on the departures from long-standing pressure delivery algorithms.
Sidebar: Details on CPAP Algorithms
Here’s a quick guide to new and emerging algorithms.
Company: NovaResp
Algorithm: cMAP
Primary Difference: Predictive AI anticipates apneas and prevents them, allowing for ~20% lower mean pressure.
Evidence: ATS 2025 abstract: Personalization of AI-enabled PAP therapy (CMAP®) algorithm shows improvement in prediction of respiratory events and preliminary success in further reduction of therapy pressure. ATS 2026 abstract to be published in May.
Availability: Investigational; designed as firmware for existing PAP platforms.
Company: React Health
Algorithm: Smart-A
Primary Difference: Uses a 32-breath rolling average for breath-by-breath precision and a 5-day average to fine-tune auto settings, including micro pressure adjustments; offers sensitivity settings.
Evidence: bench testing (unpublished)
Availability: Available now in the G3X device (released July 2025).
Company: Resmed
Algorithm: Smart Comfort
Primary Difference: Uses machine learning to recommend individualized starting comfort settings (ramp, EPR) based on patient profile.
Evidence: Retrospective analyses of Smart Comfort versus device defaults available in FDA 510(k) K251657.
Availability: Available now in beta for AirSense 11 users; full rollout late 2026.
Company: SleepRes
Algorithm: KPAP
Primary Difference: Drops pressure for the first 3 seconds of a 4-second breath; full pressure only at end-expiration.
Evidence: Kairos positive airway pressure (KPAP) equals continuous PAP in effectiveness, and offers superior comfort for obstructive sleep apnea treatment. Sleep Med. 2024 Dec.
Availability: Available on the upcoming Kricket PAP device (expected June 2026).
References
1. Messineo L, White DP, Hete B, et al. Auto-adjusting positive airway pressure: the fine line between engineering and medicine. Sleep Breath. 2025 Jul 28;29(4):253.
2. Personalized therapy comfort settings (PTCS). FDA 510(k) K251657. 5 Dec 2025.
3. Hanafi H, Hickey M, Hasan KA, et al. 2025. Late Breaking Abstract – Enhanced PAP adherence using cMAP® for sleep apnea therapy vs APAP: A preliminary 3-month analysis. Respiratory infections and bronchiectasis.
4. Hanafi H, Hasan KA, Sinclair M, et al. Personalization of AI-enabled PAP therapy (CMAP®) algorithm shows improvement in prediction of respiratory events and preliminary success in further reduction of therapy pressure [abstract]. Am J Respir Crit Care Med. 2025;211:A4755.
5. White DP, Messineo L, Thompson E, et al. Kairos positive airway pressure (KPAP) equals continuous PAP in effectiveness, and offers superior comfort for obstructive sleep apnea treatment. Sleep Med. 2024 Dec;124:268-75.
6. Noah WH, Messineo L, Hete B, et al. Treatment-emergent central sleep apnea resolves with lower inspiratory pressure. J Clin Sleep Med. 2025 Mar 1;21(3):559-64.
