The Hidden Respiratory Events Your Flow Data Isn't Showing You

To understand what's being missed, it helps to understand what current tools detect. AHI counts complete breathing stops (apneas) and sustained airflow reductions (hypopneas) that last at least 10 seconds. RERA detection looks for sequences of 3-15 breaths where effort progressively increases. NED (Negative Effort Dependence) measures the shape of each breath to detect mid-inspiratory flow drops.

These are all shape-based analyses. They look at how each breath looks. But there's a whole category of events that changes breath amplitude without changing breath shape. The airway briefly narrows or collapses for just 1-2 breaths, flow drops by 40% or more, then immediately recovers. The breath shape during the event can look perfectly normal — it's just smaller.

A brief obstruction is exactly what it sounds like: a momentary narrowing or collapse of the upper airway that lasts just 1-2 breaths. Flow amplitude drops by more than 40% from the rolling baseline, then recovers immediately. The entire event is over in 3-6 seconds — well below the 10-second threshold required for standard hypopnea scoring.

These events are individually minor. But when they happen 5-9 times per hour, they create a cumulative burden that explains the gap between your flow metrics and your oximetry data. Each brief collapse can trigger a micro-arousal, an oxygen dip, a heart rate surge — all the physiological responses that fragment sleep and drive next-day symptoms.

NED measures Negative Effort Dependence: the ratio of peak inspiratory flow to flow at the midpoint of inspiration. When the airway is progressively narrowing during a breath, mid-inspiratory flow drops relative to peak flow, producing a high NED value. This is the classic "scooped out" flow shape that indicates flow limitation.

But brief obstructions work differently. The airway doesn't gradually narrow during inspiration — it snaps partially closed for the entire breath, reducing overall amplitude while maintaining a normal waveform shape. Peak and mid-inspiratory flow both drop proportionally. NED stays normal. The Flatness Index stays normal. The Glasgow Index components stay normal. By every shape metric, the breath looks fine. It's just 40-60% smaller than it should be.

This is why we call these events "NED-invisible"— they represent real airway compromise that shape-based flow analysis cannot detect. In our analysis, the majority of brief obstructions have NED values well below the 34% flow limitation threshold.

Brief obstructions are individual events, but there's also value in looking at how stable your breathing amplitude is across the entire night. Normal tidal breathing has natural variability — your breaths aren't all exactly the same size. But that variability typically falls within a predictable range (coefficient of variation around 15-20%).

When the airway is intermittently compromising, breath amplitude becomes erratic. Some breaths are normal, some are reduced, some are recovery breaths that overshoot. The coefficient of variation climbs. By dividing the night into 5-minute epochs and measuring amplitude variability within each epoch, you can see when and how often the airway is behaving unstably — even if individual breath shapes look fine.

If your standard flow metrics look good but you're still symptomatic, brief obstructions and amplitude instability are worth investigating. Here's what to look for and discuss with your sleep physician:

AirwayLab v1.2 introduces amplitude-based detection alongside the existing shape-based analysis. The detector tracks peak inspiratory flow (Qpeak) against a rolling 30-breath median baseline. When a breath's Qpeak drops more than 40% below that baseline, it's flagged as a brief obstruction.

Each detected event is also checked for NED visibility — whether the NED shape analysis would have caught it independently. Events where NED is below 34% during the amplitude drop are flagged as NED-invisible, giving you a clear picture of what standard shape analysis is missing.

For sustained flow reductions (≥30% drop lasting ≥10 seconds), AirwayLab reports a Hypopnea Index. When your ResMed machine provides its own hypopnea count via EVE.edf, AirwayLab uses the machine's number — it has access to internal pressure and flow algorithms that external analysis can't replicate. When EVE data isn't available, AirwayLab falls back to its own amplitude-based detection. Either way, you see a single unified number.

The new Airway Stability section in the Flow Analysis tab shows Brief Obstruction Index, Hypopnea Index, and Amplitude CV with traffic light indicators, trend arrows, and first-half/second-half comparisons — the same analysis patterns you're used to from the existing NED and Glasgow sections.