Definition, intended role, and neighboring categories
Adaptive servo-ventilation, commonly abbreviated ASV, is a noninvasive positive-pressure technology used principally for central sleep apnea, periodic breathing, and selected central events that emerge during therapy. It continuously observes recent breathing, changes pressure support from breath to breath, and supplies automated backup breaths when needed to damp oscillation between overventilation and underventilation. It is a prescription medical device, not a marketing synonym for ordinary auto-adjusting CPAP.
ASV uses expiratory pressure to help maintain upper-airway patency and the difference between inspiratory and expiratory pressure to create rapidly variable pressure support. Auto-bilevel devices primarily track obstruction, bilevel devices with a backup rate support breathing according to preset timing, and volume-assured pressure support moves toward a clinician-selected ventilation target. Their targets, time scales, and risks differ, so they are not interchangeable merely because each can change pressure automatically.
Adaptive servo-ventilation in sleep medicine must also be distinguished from similarly abbreviated modes such as adaptive support ventilation in critical care. Most home ASV products are non-life-support devices. They do not replace emergency resuscitation, a critical care ventilator, or a home life-support system designed for a ventilator-dependent person. The authorized age, condition, setting, and interface must be confirmed in the local labeling for the specific product.
Sensing, targets, and breath-by-breath control
The device measures total flow in the mask circuit and estimates patient flow, leak, respiratory phase, and the recent ventilation pattern. Some algorithms reference a moving level of minute ventilation, whereas others use inspiratory peak flow or a different time window. Support rises when ventilation wanes, falls when breathing is excessive, and backup timing is selected by the algorithm. The aim is dynamic stabilization, not forcing every breath to the same tidal volume.
Expiratory pressure may be fixed or, in an authorized automatic mode, may respond to obstructive events. Pressure support changes more rapidly within prescribed bounds to address central weakening, pauses, or periodic fluctuation. ASV can therefore address both central and obstructive components, but these are separate control tasks. Higher expiratory pressure does not directly restore absent respiratory drive, and more pressure support does not prove that the upper airway is open.
Because the algorithm uses recent breathing as its reference, awake sighs, talking, turning, breath holding, rapid sleep-onset changes, a large mask leak, or the wrong interface can distort its estimate and response. Manufacturers and software versions differ in their definitions of baseline, target, leak compensation, backup timing, and rate of pressure change. A shared mode name does not establish equivalent waveforms, patient experience, or clinical outcomes.
Diagnosis, cause, and candidate selection
Before ASV is selected, a technically adequate sleep study and clinical information distinguish central, obstructive, mixed, and treatment-emergent events and determine whether they occurred during sleep or wakefulness. Identification of a central event requires information such as respiratory effort. A flow label generated by the device, or data from a wearable, cannot establish the causal diagnosis by itself. Brief sleep-onset central pauses should not automatically be pathologized.
Central sleep apnea can occur with heart failure, medication or substance exposure, neurologic or other medical disorders, high altitude, or primary ventilatory instability. It may also emerge after PAP is started for obstructive sleep apnea. Evaluation considers history, symptoms, cardiac, pulmonary and neurologic status, exposures, sleep-study findings, and gas-exchange data when needed. Treatable contributing factors are optimized rather than focusing only on eliminating an event count.
Current professional guidance conditionally suggests considering ASV for several adult central sleep apnea etiologies, but the certainty of evidence is limited. Decisions center on symptoms, sleep, and quality of life that matter to the patient and use shared decision-making to compare applicable options such as CPAP, bilevel therapy with a backup rate, oxygen, medication, or phrenic nerve stimulation. A guideline suggestion does not mean that every patient or every model is authorized for use.
Heart-failure evidence and safety boundaries
In a specific trial population with chronic symptomatic heart failure, reduced left ventricular ejection fraction, and predominantly central events, a large study found that one ASV approach did not improve the primary composite cardiovascular outcome and was associated with a signal of increased cardiovascular mortality. The result led to explicit contraindications or warnings for corresponding populations in some products and regions. It also demonstrated that suppressing respiratory events does not establish improved heart-failure outcomes.
Later trials using different algorithms and designs did not reproduce the same mortality signal, but they likewise did not establish ASV as a general treatment for improving heart-failure survival or preventing cardiovascular events. The differing results do not simply cancel one another, and evidence for one device cannot be generalized to every servo algorithm. Population, ejection fraction, event mixture, device logic, adherence, and outcome definitions all require examination.
Before treatment, clinicians identify heart-failure symptoms and phenotype, obtain information adequate to judge current left ventricular function, and check current local guidance, regulatory communications, and model-specific labeling. For heart failure with reduced ejection fraction, current guidance calls for more cautious shared decision-making in experienced centers with close monitoring and follow-up. Any product contraindication that remains in force takes precedence and cannot be waived by an encyclopedia entry or remote platform.
Titration, interfaces, and data interpretation
A qualified team titrates therapy during valid sleep with adequate signal quality, observing obstruction, central oscillation, oxygenation, arousals, leak, and synchrony before establishing expiratory-pressure ranges, pressure-support bounds, and backup behavior. Bounds that are too narrow may fail to stabilize breathing; an excessively broad or poorly matched response may cause abrupt pressure, leak, arousal, or a sensation of overventilation. Another person's settings are not copied, and self-titration is unsafe.
Mask type and tension, mouth leak, tubing, humidifier, filter media, and supplemental-oxygen connection all influence flow estimation and therapy. A large leak can cause ineffective support, false triggering, or unnecessary pressure swings, while excessive tightening can injure the skin. Supplemental oxygen is used only when prescribed and through an approved connection, with fire precautions appropriate to an oxygen-enriched environment.
Follow-up data may include usage, residual event indices, percentage of periodic breathing, leak, expiratory pressure, pressure support, and backup-breath measures, but definitions and valid-data rules vary by device. A device does not directly observe all sleep, respiratory-effort, and arousal signals available in full polysomnography. Trends are combined with symptoms, detailed waveforms, oximetry, or other testing; treatment is not managed solely to produce the lowest AHI.
Outcome review, risks, and escalation
Review asks whether sleep, daytime function, breathing discomfort, or quality of life actually improved and also examines use coverage, mask tolerance, leak, and residual events. Persistent central events prompt renewed assessment of cause, medication, heart-failure status, altitude, data quality, and mode selection rather than unlimited increases in support. A numerical change without symptoms and a patient-valued clinical benefit are reported separately.
Common burdens include oral or nasal dryness, pressure marks, conjunctival irritation, congestion, aerophagia, bloating, leak, noise, claustrophobia, arousal, and dyssynchrony with rapid pressure changes. Positive pressure can also alter hemodynamics. Misclassification or an unsuitable mode can delay correct management of hypoventilation, lung disease, neuromuscular disease, or heart failure. ASV neither discontinues a causative medication nor replaces treatment of the underlying condition.
New or worsening breathlessness at rest, chest pain, syncope, cyanosis, altered consciousness, persistent severe hypoxemia, or signs of acute heart failure require prompt emergency assessment rather than waiting for a cloud-data review. The alternative plan for device failure, power loss, or intolerance must match the person's dependence on ventilatory support. A non-life-support ASV device must not be treated as the sole rescue system.
Cleaning, connectivity, and lifecycle governance
The mask, cushion, headgear, tubing, water chamber, and filter media are cleaned, dried, and replaced according to their respective instructions; the main unit is not immersed or opened for improvised repair. Ozone, ultraviolet devices, fragrance oils, harsh chemicals, stacked filters, and unapproved adapters are avoided unless specifically validated by the manufacturer. They can damage materials, leave irritants, raise resistance, or corrupt flow estimates.
A remote platform can assist with adherence and trend review, but proposing settings, approving a change, and actually sending parameters require tiered permissions, identity verification, audit records, and rollback. During network loss, the device retains validated local therapy. Reconnection must not let an old prescription overwrite a newer one, bind data to the wrong patient, or shift timestamps. After an algorithm, firmware, or data-field update, historical comparability is checked.
Procurement and long-term management examine the authorized population, algorithm description, pressure and leak performance, alarms and power capability, data access, cybersecurity, consumables, service, recalls, and migration at end of support. Model, serial number, firmware, prescription source, mask and accessory configuration, and material changes are retained. Claims of intelligent learning, rapid response, or event elimination are not independent evidence of safety, survival benefit, or universal effectiveness.