info@cpapstudio.co.uk
+44 7899 539620
CPAP technology is evolving faster than most patients realise. From real-time AI coaching to NHS remote monitoring programmes, here is what is already here, what is coming, and what it means for your therapy.
From the SD Card to the Cloud: How CPAP Data Has Transformed
Until relatively recently, accessing your CPAP therapy data as a patient meant either waiting for a clinic appointment where a clinician would read an SD card from your device, or purchasing specialised third-party software to read the card yourself. The data was there, but it was largely inaccessible in any practical sense to the person whose sleep it described.
The shift to cloud-connected CPAP devices over the past decade has fundamentally changed this relationship between patient and data. Modern ResMed AirSense and Philips DreamStation devices transmit nightly therapy data automatically over a cellular or Wi-Fi connection to cloud platforms that are simultaneously accessible to the patient via a smartphone app and to their clinical team via a clinician-facing portal. The same night’s data that a patient sees in their MyAir app the following morning is visible in the ResMed AirView platform to their sleep clinic without either party having to do anything beyond the routine of using the device.
Data stored locally on SD card. Access required clinical appointment or patient-purchased software (e.g. SleepyHead / OSCAR). No real-time visibility for either clinician or patient.
ResMed AirSense 10 introduced cellular data transmission. Clinicians could review data remotely via AirView. Patient-facing apps began appearing. Still largely a clinical tool rather than a patient engagement tool.
ResMed MyAir launched publicly, delivering nightly scores and encouragement directly to patients. Evidence began emerging that app-based engagement improved adherence. Philips DreamMapper followed. Patient-facing data became a mainstream expectation.
NHS services began piloting structured remote monitoring programmes triggered by cloud data alerts. Machine learning models applied to CPAP datasets began identifying early drop-off risk and mask fit issues. AirSense 11 launched with enhanced AutoSet algorithms and refined app integration.
AI-generated personalised coaching delivered via apps. Predictive models identifying patients at risk of abandonment before they disengage. NHS-commissioned remote monitoring pathways scaling in several regions. Third-party integration platforms expanding patient data access and interoperability.
Patient Apps: What They Already Do and Where They Are Heading
Patient-facing CPAP apps have moved considerably beyond their initial role as simple data dashboards. The current generation of apps led by ResMed MyAir combines data visualisation with behavioural engagement features, and is increasingly incorporating elements of personalised coaching driven by pattern recognition algorithms applied to each user’s own therapy history.
- Daily score 0–100 with sub-scores
- AHI, leak rate, usage hours
- Weekly trends and streaks
- AI-generated coaching messages
- Mask-fit troubleshooting prompts
- Connected to AirSense 10 & 11
- Usage, AHI, leak data
- Goal-setting features
- Therapy trends over time
- Compatible with DreamStation 1 & 2
- Clinician data sharing option
- Available iOS and Android
- Detailed waveform data
- Breath-by-breath analysis
- Event type breakdown
- Flow limitation visualisation
- Pressure graph night-by-night
- Free; requires SD card or USB
How AI Coaching Features Currently Work in MyAir
ResMed’s MyAir app now includes what the company describes as an AI coaching feature that delivers personalised in-app messages based on patterns in each user’s therapy data. Unlike generic advice, these messages are intended to be relevant to the user’s specific recent experience for example, noting a pattern of mask leak at a particular time of night and suggesting specific corrective steps, or recognising that a user who previously had good adherence has skipped three nights in a row and delivering a re-engagement message tailored to the gap.
The underlying mechanism draws on machine learning models trained on the large datasets generated by millions of connected CPAP users globally, identifying patterns that are predictive of specific problems or behaviours and mapping them to appropriate coaching responses. The clinical evidence base for AI coaching in CPAP therapy is still accumulating, but early studies comparing app-supported to app-unsupported CPAP therapy have generally shown improved adherence outcomes in the supported group.
Remote Monitoring: What the NHS Is Already Doing
Remote monitoring of CPAP patients using the cloud-transmitted data from connected devices to manage clinical caseloads without requiring routine face-to-face appointments has moved from research pilot to mainstream NHS practice in several regions, and is likely to expand significantly over the next few years driven by the twin pressures of growing sleep apnoea prevalence and constrained clinic capacity.
Rather than reviewing every patient’s data every month, clinical teams configure threshold alerts in platforms such as ResMed AirView. When a patient’s AHI exceeds a defined level, their leak rate consistently surpasses a threshold, or their usage hours fall below a minimum for a specified number of consecutive nights, the system automatically flags their record for proactive clinical contact.
This exception-based approach allows a single sleep specialist nurse to monitor hundreds of patients simultaneously, intervening only where the data indicates a problem rather than spending clinic time on patients whose therapy is performing well. For patients, this means faster intervention when something goes wrong typically within days rather than at the next scheduled appointment months away.
Beyond simple threshold alerts, machine learning models are being developed and trialled that predict which CPAP patients are at highest risk of therapy abandonment before they actually disengage using subtle patterns in their early therapy data that are not obvious from the headline AHI and usage metrics alone.
These risk scores allow clinical teams to direct their most intensive support proactively toward the patients most likely to struggle, rather than reactively toward those who have already stopped. Early research in this area, including work drawing on ResMed’s global device dataset, has shown promising accuracy in identifying high-risk patients within the first two to four weeks of therapy.
NHS sleep services rapidly expanded telephone and video CPAP review clinics during the pandemic, and many have retained these pathways for straightforward routine reviews where the clinician can access data remotely. A video consultation combined with clinician access to the patient’s full AirView or equivalent data can replicate the majority of what a routine face-to-face review covers, while reducing travel burden for patients and venue costs for services.
Most sleep services now operate a hybrid model virtual for routine review with good data, face-to-face for new patients, complex cases, and mask fittings where a physical assessment is genuinely necessary.
Integrated Care Systems in England are beginning to explore how CPAP monitoring data can be shared across healthcare settings for example, making a patient’s therapy adherence data visible to their GP or cardiologist alongside other health records. This kind of cross-setting integration addresses the longstanding problem of sleep apnoea management existing in a clinical silo, disconnected from cardiovascular, metabolic, and mental health care pathways where OSA is clinically relevant.
True data integration at this level faces significant interoperability and governance challenges, and is likely to be a medium-term rather than immediate development for most NHS regions.
What Smart CPAP Devices Can Already Detect Automatically
Modern AutoCPAP algorithms are considerably more sophisticated than the first-generation auto-titrating devices of the early 2000s. Current devices from ResMed and Philips incorporate algorithms that classify breathing events in real time, adapt pressure in response to detected obstruction, and in some cases discriminate between event types with meaningful clinical relevance.
| Detection Capability | How the Device Responds | Clinical Significance |
|---|---|---|
| Obstructive apnoea | Increases pressure to reopen airway | Core function — well established |
| Hypopnoea | Increases pressure proportionally | Included in AHI calculation |
| Flow limitation (snore/RERA) | Pre-emptively raises pressure before full obstruction | Reduces upper airway resistance — emerging evidence |
| Large mask leak | Flags in data; some devices adjust response | Triggers app alert; clinical review may be needed |
| Treatment-emergent central apnoeas | Flagged in data; some devices switch response algorithm | Requires clinical review if persistent |
| Cheyne-Stokes respiration pattern | ResMed VAuto / ASV devices detect and respond specifically | Indicates need for BiPAP/ASV — clinical escalation |
| Estimated sleep position | Some devices log positional data via accelerometer | Identifies positional OSA — emerging clinical use |
On the Horizon: What CPAP Technology Is Working Toward
The following developments are credibly in progress across the sleep medicine technology landscape in 2026, based on published research, company announcements, and known clinical trial activity. Each comes with meaningful caveats about the gap between what is technically possible and what reaches routine UK clinical practice.
- Contactless sleep monitoring integration: Several companies are developing non-wearable radar and radiofrequency sensors that monitor sleep staging, breathing, and movement without any physical contact. Integration of this data with CPAP device data could provide a much richer picture of sleep architecture than CPAP devices can currently capture alone potentially allowing monitoring of whether CPAP is restoring normal sleep staging rather than simply reducing AHI.
- Fully automated pressure optimisation: Current AutoCPAP devices adjust within a prescriber-set range but cannot change their own range or mode without clinical intervention. Research is underway toward closed-loop systems that could more autonomously identify and adjust to changing patient needs though the regulatory and safety challenges of autonomous clinical adjustment are considerable.
- Wearable biomarker integration: Smartwatch and wearable device data including heart rate variability, SpO₂, and sleep stage estimates are increasingly being explored as complementary data streams alongside CPAP device telemetry, allowing a more complete picture of cardiovascular and sleep health to inform therapy decisions.
- Predictive maintenance and supply: Using therapy data patterns to automatically trigger replacement cushion or hose delivery before a component fails, rather than relying on patient-initiated replacement requests, is a relatively near-term application of existing data infrastructure that could meaningfully reduce the therapy gap caused by delayed component replacement.
- Population-level AI insights and personalised pressure optimisation: The datasets generated by millions of connected CPAP devices globally represent an extraordinary resource for identifying population-level patterns in OSA severity, treatment response, and cardiovascular outcomes. Research drawing on these datasets is already producing insights that are beginning to influence clinical guideline development.
What This Means for UK CPAP Patients Today
For a patient currently on CPAP in the UK, the practical upshot of the technology developments described above is a set of concrete actions worth taking now some of which are already available and simply underused.
- Download and actively use your device's patient app. If you use a ResMed AirSense device and have not set up MyAir, this is the single most immediate step you can take. The coaching features, trend data, and in-app guidance available are considerably more useful than most patients realise, and are already using AI-driven personalisation based on your own therapy data.
- Ask your sleep clinic whether they use remote monitoring. Many NHS services now actively monitor patient data via AirView or equivalent platforms. If yours does, confirm that your device is enrolled and that your data is reaching the monitoring team. If they do not yet have a formal remote monitoring programme, knowing this helps you understand what level of active oversight your therapy receives between appointments.
- Explore OSCAR if you want deeper data access. OSCAR (Oscar's CPAP Reporting Software) is a free, open-source application that provides far more detailed breath-by-breath data than any manufacturer app, including waveform visualisation and event type breakdown. It requires SD card or USB data access rather than cloud data, but for technically interested patients it provides a level of insight that no consumer app currently matches.
- Be aware of your consent to data sharing. When you enrol your device in a cloud platform, your anonymised therapy data may contribute to the large datasets used to develop the AI models described in this article. This is generally covered in device manufacturer terms of service, and the aggregate datasets are subject to data protection regulation but it is worth being a conscious participant in this data relationship rather than an unaware one.
