Frequently Asked Questions

When patients are on a breathing machine, two forces work together to inflate the lungs: the machine pushes air in from above, while the patient’s own breathing muscles pull from below by expanding the chest. Together, they can stretch the lungs far more than either force alone. The problem is that breathing machines only show the pressure they deliver — they cannot detect how hard the patient is pulling. When the patient breathes actively, the actual stress on the lungs can be far greater than what the machine’s screen suggests. Think of it like an iceberg: the machine only shows the tip above water, while the patient’s hidden effort — often the larger portion — remains invisible.

This hidden stress matters because excessive lung stretching is a major cause of further lung damage. Clinical studies have confirmed that the total lung stress during active breathing is linked to patient survival — but clinicians cannot assess it without knowing the patient’s effort. Even the most basic safety check — the “plateau pressure” that doctors routinely measure when the machine does all the breathing — becomes unreliable the moment the patient starts breathing on their own. At the same time, the level of effort itself also needs to be managed carefully: too much effort can injure the breathing muscles, while too little causes them to weaken rapidly. To protect both the lungs and the breathing muscles simultaneously, doctors need continuous information about how hard the patient is breathing — and right now, standard breathing machines simply cannot provide it.

Doctors can measure a patient’s breathing effort by inserting a thin tube with a small balloon down the throat into the food pipe (esophagus). The pressure changes in this balloon closely mirror the pressure inside the chest cavity. However, the main obstacle is not the tube insertion itself — it is how difficult it is to get reliable measurements afterward. The balloon must be placed in exactly the right position and checked with special tests. Its air volume needs to be adjusted carefully and re-checked every few hours because of slow air leaks. The signal is affected by the heartbeat and by contractions of the food pipe itself. Determining exactly when the patient starts and stops each breathing effort can be surprisingly difficult. On top of all this, measuring the stiffness of the chest wall — which is needed to calculate the actual breathing effort — requires the patient to be deeply sedated and not breathing on their own.

The catheters are also expensive, the procedure takes at least 30 minutes, and it cannot be used in patients with certain medical conditions. Because of all these barriers, very few ICUs use this technique routinely during active breathing, even though the information it provides is extremely valuable.

There are several ways to estimate breathing effort without an invasive catheter, but none provides continuous, reliable information.

P0.1 captures only the first tenth of a second of effort — kept deliberately brief so the patient doesn’t notice. It reflects the brain’s drive to breathe, not how hard the patient actually breathes throughout the breath.

Expiratory occlusion (ΔPocc) blocks the airway for one breath and uses a fixed factor to convert the measured pressure into an effort estimate. But how patients react to the blockage varies with their alertness and sedation, so a fixed factor is inherently imprecise. Its developers describe it as a screening test, not a monitoring tool.

End-inspiratory hold pauses the machine at the end of a breath-in to estimate effort from the plateau pressure. It only works if the patient fully relaxes during the pause — which often doesn’t happen.

Mathematical shape fitting reconstructs an effort waveform by assuming it follows a predetermined mathematical curve. When the real effort doesn’t match that shape, the result looks plausible but is systematically wrong.

None of these can continuously show how hard the patient is breathing throughout each breath, across different machine settings, especially when the patient and machine are out of sync.

CVent is an AI-powered technology that solves a problem conventional breathing machines cannot: it calculates both how hard the patient is breathing and how stiff the patient’s lungs are, in real time, using only the standard signals every breathing machine already measures. No additional catheters, sensors, or maneuvers are needed. From this, it can continuously estimate the key safety indicators for lung protection — including the plateau pressure, the total driving pressure, and the lung stress contributed by the patient’s own effort — for every single breath, without ever pausing the machine.

What makes it different from all other non-invasive methods: it shows the patient’s effort as a complete, continuous waveform — not just a single number or a brief snapshot. It can estimate the plateau pressure and driving pressure for every breath without pausing the machine. Traditional mathematical methods work like memorized answers to predicted exam questions — they only work when the real situation matches what was expected. CVent, by contrast, is like a student who has truly mastered the subject by studying millions of examples: it has learned the underlying patterns of how breathing works, so it can handle unfamiliar and complicated situations that were never explicitly programmed for. It works across all common machine settings. It remains accurate even when the patient and machine are out of sync — and because it shows exactly when the patient is trying to breathe in and out, it makes timing mismatches between the patient and the machine immediately visible to the clinician. And unlike fixed algorithms that never change, CVent grows stronger as more patient data becomes available — becoming more accurate and more adaptable over time.

The basic challenge is straightforward: when a patient breathes on a machine, the machine knows the pressure it delivers and the airflow it measures, but these two pieces of information alone are not mathematically sufficient to determine the patient’s effort — there are too many unknowns. CVent is a proprietary software system that overcomes this using artificial intelligence trained on more than one million breathing patterns, generated by both computer simulations and physical lung simulators. Through this extensive training, it learned to reverse-engineer the patient’s breathing effort from the two signals the machine already measures. When applied to a real patient, it needs about ten breaths to learn that patient’s specific characteristics, and then it continuously calculates and displays the effort waveform in real time.

CVent’s most important clinical value is that it gives doctors the information they need to protect both the lungs and the breathing muscles at the same time — something that has been impossible without invasive monitoring.

First, it restores a key safety measurement that is lost when the patient starts breathing on their own. During fully machine-controlled breathing, doctors can easily check the “plateau pressure” — the most basic indicator of whether the lungs are being overstretched. But the moment the patient starts breathing actively, this measurement becomes unreliable. CVent solves this by estimating plateau pressure for every breath without any interruption. It also reveals the true total stress on the lungs — including the hidden portion contributed by the patient’s own effort — so doctors can judge whether ventilation is truly safe.

Second, it helps protect the breathing muscles. Too much effort injures them; too little causes them to weaken rapidly. CVent continuously shows how hard the patient is working, so doctors can adjust the machine’s support level and medications to keep effort in a healthy range.

Third, it helps decide when the patient is ready to come off the machine. By continuously tracking muscle strength and workload over time, CVent can help identify when a patient is strong enough to breathe on their own — and when they are not.