Continuous Respiratory Surveillance
Part of the Clarion Sentinel Detection Suite

Every breath is a data point

From the first wheeze of a COPD exacerbation to the final decision to extubate — Sentinel Respira monitors, predicts, and guides every stage of respiratory disease with intelligence no single clinician can match.

3.4%
Annual increase in respiratory failure mortality — over 1.4 million deaths between 2014-2018 in the US alone. Sentinel Respira intervenes earlier.
3.23M
COPD deaths worldwide per year — 3rd leading cause
3M+
US hospitalizations for respiratory failure annually
15-20%
Extubation failure rate — each attempt carries risk
0.80+
AI prediction AUC for respiratory outcomes
The Respiratory Crisis

Breathing is the most fundamental act of being alive — and respiratory failure is the most common reason patients end up in intensive care. COPD alone kills 3.23 million people annually and is the third leading cause of death worldwide. Asthma exacerbations fill emergency departments. Pulmonary fibrosis progresses silently. Ventilator-associated complications kill patients we are trying to save. And at every stage, the decisions — when to intubate, how to ventilate, when to wean — are made with insufficient data, subjective judgment, and tools that have not fundamentally changed in decades.

Sentinel Respira is the first AI platform that treats respiratory disease as a continuous spectrum rather than a series of disconnected clinical decisions. It monitors pulmonary function longitudinally, predicts exacerbations before they occur, optimizes ventilator management in real time, determines the precise moment a patient can safely be liberated from mechanical ventilation, and surveils post-extubation recovery to prevent reintubation. One platform. Eight engines. Every breath accounted for.

The Respiratory Failure Continuum

From chronic disease to acute crisis — and back

Respiratory failure is not a single event. It is a continuum with distinct intervention windows at every stage.

1
Chronic Disease & Subclinical Decline
COPD, asthma, pulmonary fibrosis, and bronchiectasis progress silently. Pulmonary function declines at rates imperceptible visit-to-visit. Exacerbation triggers accumulate — environmental exposures, medication non-adherence, viral infections — while the patient feels "fine." This is where longitudinal AI monitoring detects the trajectory no clinic visit can capture.
2
Pre-Exacerbation Window
Sentinel detects patterns here — 24-72 hours before symptoms
Subtle shifts in respiratory rate, SpO2 baseline, activity levels, and inhaler use patterns signal an impending exacerbation. Wearable sensors and smart inhalers detect changes invisible to the patient. This is the window where outpatient intervention — steroids, antibiotics, bronchodilator escalation — can prevent hospitalization entirely.
3
Acute Exacerbation & ED Presentation
Dyspnea, wheezing, hypoxemia, hypercapnia. The patient presents to the emergency department. The critical decision: can this be managed with non-invasive ventilation (NIV), or does this patient need intubation? Incorrect decisions in both directions carry serious consequences — premature intubation subjects the patient to ventilator risks; delayed intubation allows further decompensation.
4
Mechanical Ventilation
The patient is intubated and placed on mechanical ventilation. Every hour on the ventilator increases the risk of ventilator-associated pneumonia, barotrauma, ventilator-induced lung injury, ICU-acquired weakness, and delirium. The ventilator saves lives — and the ventilator destroys lives. The art is in the settings, the timing, and the liberation.
Each additional ventilator day increases complications and mortality
5
Weaning & Extubation
15-20% extubation failure rate — each failure doubles mortality
The most consequential decision in respiratory critical care: when to remove the tube. Extubate too early and the patient reintubates — which doubles mortality. Extubate too late and the patient suffers unnecessary ventilator days. Current tools (RSBI, SBT) have insufficient sensitivity and specificity. AI models integrating multidimensional data are transforming this decision.
Failed extubation doubles ICU mortality
6
Post-Extubation & Readmission Risk
The 72 hours after extubation are the highest-risk period for respiratory recompensation. And after discharge, the 30-day readmission window determines whether the patient cycles back through the entire cascade. Sentinel Respira monitors both — the acute post-extubation phase and the chronic post-discharge trajectory — to break the cycle of crisis, recovery, and relapse.
Detection & Management Engines

Eight engines. Every breath monitored.

From outpatient chronic disease management through ICU ventilator optimization to post-discharge surveillance.

Engine 01
COPD Exacerbation Prediction
Predicts acute exacerbations 24-72 hours before symptom onset using wearable data, inhaler telemetry, and environmental factors.
Prevents 44% of COPD hospitalizations through early outpatient intervention

COPD exacerbations are the primary driver of disease progression, hospitalization, and death. Each severe exacerbation accelerates lung function decline and increases the risk of subsequent episodes — a vicious cycle. Sentinel Respira breaks this cycle by predicting exacerbations 24-72 hours before symptoms manifest, using pulse oximetry trends, respiratory rate changes from wearable devices, smart inhaler usage patterns (rescue inhaler frequency, technique quality), air quality data from environmental sensors, sleep quality metrics, and activity level changes. When the risk score crosses a threshold, the system alerts the patient's pulmonologist and recommends outpatient escalation — oral steroids, antibiotic initiation, bronchodilator intensification — preventing the emergency department visit entirely.

Performance
24-72hr
Prediction window before exacerbation symptom onset
44%
Reduction in COPD-related hospitalizations
0.804
AUC for acute respiratory failure prediction in COPD
Input Signals
SpO2 Baseline DriftRescue Inhaler FrequencyRespiratory RateActivity LevelSleep QualityAir Quality IndexTemperatureHumidity
Engine 02
Acute Respiratory Failure Detection
Continuous monitoring of oxygenation, ventilation, and work-of-breathing to detect respiratory decompensation hours earlier.
Detects respiratory failure trajectory 4.2 hours before clinical deterioration

Acute respiratory failure does not strike suddenly — it builds. SpO2 drifts downward. Respiratory rate creeps upward. PaO2/FiO2 ratios narrow. Work of breathing increases as accessory muscles engage. Sentinel Respira monitors these trajectories continuously, identifying the pattern of impending respiratory failure hours before a crisis triggers a code or rapid response call. For hospitalized patients, the system integrates vital signs, arterial blood gas trends, chest imaging evolution, and clinical context to differentiate between respiratory failure from pneumonia, heart failure, PE, ARDS, and neuromuscular causes — enabling targeted intervention rather than generic supportive care.

Performance
4.2hr
Earlier detection of respiratory failure trajectory
91%
Accuracy in differentiating respiratory failure etiology
38%
Reduction in unplanned intubations through earlier intervention
Engine 03
NIV Outcome Prediction
Predicts non-invasive ventilation success or failure within the first hour — guiding the intubation decision.
Determines within 60 minutes whether NIV will succeed or the patient needs intubation

Non-invasive ventilation is a cornerstone of acute respiratory failure management — but it fails in a significant proportion of patients. Delayed recognition of NIV failure is one of the most dangerous errors in respiratory medicine, as it subjects the patient to prolonged respiratory distress while delaying definitive airway management. Sentinel Respira analyzes vital signs, respiratory mechanics, blood gas trajectories, and clinical response within the first 60 minutes of NIV initiation to predict whether the patient will succeed on NIV or require intubation — enabling the clinical team to prepare for escalation before crisis-level deterioration occurs.

Performance
60min
Prediction of NIV outcome from initiation data
84%
Accuracy predicting NIV failure requiring intubation
2.8hr
Earlier escalation to intubation when NIV fails
Engine 04
Ventilator Optimization Intelligence
Real-time optimization of ventilator settings — PEEP, tidal volume, FiO2, respiratory rate — individualized to each patient.
Reduces ventilator-induced lung injury by 28% through continuous parameter optimization

Mechanical ventilation is a double-edged sword. While it sustains life, inappropriate settings — excessive tidal volumes, inadequate or excessive PEEP, unnecessary FiO2 levels — directly cause ventilator-induced lung injury, barotrauma, and oxygen toxicity. Sentinel Respira continuously optimizes ventilator parameters based on real-time lung mechanics (compliance, resistance, driving pressure), oxygenation targets, CO2 clearance needs, and patient-ventilator interaction patterns. The system detects patient-ventilator asynchrony (double-triggering, auto-triggering, flow starvation) and recommends setting adjustments before asynchrony causes patient distress or injury.

Performance
28%
Reduction in ventilator-induced lung injury
92%
Patient-ventilator asynchrony detection accuracy
1.8day
Average reduction in ventilator days
Engine 05
Weaning & Extubation Prediction
AI-driven extubation readiness assessment that outperforms RSBI and SBT — predicting success with 88% accuracy.
The most consequential decision in respiratory care — AI gets it right 88% of the time

Extubation failure occurs in 15-20% of patients, and each failed extubation doubles ICU mortality. Current tools — the rapid-shallow breathing index (RSBI) and spontaneous breathing trial (SBT) — have insufficient sensitivity and specificity, leaving clinicians guessing on the most high-stakes decision in respiratory medicine. Sentinel Respira integrates multidimensional data — respiratory mechanics during SBT, cough strength, secretion volume, mental status, hemodynamic stability, nutritional status, and diaphragm function (via ultrasound when available) — to predict extubation success with 88% accuracy, significantly outperforming any single traditional metric.

Performance
88%
Extubation success prediction accuracy
42%
Reduction in failed extubations at deployed sites
1.4day
Earlier safe extubation in patients kept ventilated unnecessarily
Engine 06
Pulmonary Function Trend Intelligence
Longitudinal analysis of spirometry, DLCO, and lung volumes — detecting disease progression invisible in single-visit data.
Identifies accelerated lung function decline 6-12 months before clinical concern

A single pulmonary function test is a snapshot. A series of PFTs over years is a trajectory — and that trajectory reveals disease progression, treatment response, and impending exacerbation risk in ways no single visit can capture. Sentinel Respira analyzes every historical FEV1, FVC, FEV1/FVC ratio, DLCO, and lung volume measurement, calculates the rate of decline for each parameter, compares it to age-adjusted population norms, and identifies patients whose decline is accelerating — triggering treatment intensification, specialist referral, or transplant evaluation before the patient reaches a crisis that could have been prevented.

Performance
6-12mo
Earlier detection of accelerated decline vs. standard clinic review
34%
More patients referred for transplant evaluation at appropriate timing
Engine 07
Asthma Crisis Detection & Management
Real-time asthma severity classification and exacerbation prediction using smart inhaler data and environmental triggers.
Reduces ED visits for asthma by 38% through early outpatient escalation

Asthma kills 455,000 people annually worldwide — and the vast majority of asthma deaths are preventable. Sentinel Respira monitors asthma patients continuously through smart inhaler telemetry (rescue inhaler frequency, technique scores, controller medication adherence), wearable vital sign data, environmental trigger exposure (pollen counts, air quality, temperature changes), and peak flow trends. When the system detects the pattern of an impending exacerbation, it alerts both the patient and their pulmonologist, recommending controller medication escalation, oral steroid initiation, or clinic evaluation — preventing the emergency department spiral that drives asthma mortality.

Performance
38%
Reduction in asthma-related ED visits
24hr
Average early warning before severe exacerbation
Engine 08
Post-Extubation & Readmission Surveillance
Monitors the 72-hour post-extubation window and 30-day post-discharge period to prevent reintubation and readmission.
30% of COPD patients are readmitted within 30 days — most preventably

The 72 hours after extubation are the most dangerous for respiratory recompensation — stridor, laryngeal edema, secretion retention, and diaphragmatic fatigue can all precipitate reintubation. And after discharge, 30% of COPD patients are readmitted within 30 days. Sentinel Respira monitors both windows: continuous respiratory surveillance in the post-extubation ICU phase (respiratory rate, SpO2, work of breathing, cough effectiveness) and outpatient monitoring post-discharge (symptom questionnaires, wearable vital signs, inhaler adherence, and environmental triggers) — breaking the cycle of crisis and readmission that defines chronic respiratory disease.

Performance
34%
Reduction in reintubation within 72 hours of extubation
28%
Reduction in 30-day COPD readmission rate
Proven Impact

Breathing restored. Ventilator days reduced. Readmissions prevented.

Results from our deployed health systems.

Pulmonary Medicine Network — 14 Clinics, 28,000 COPD Patients

Exacerbation prediction preventing 44% of COPD hospitalizations

The Outcome

Deployed across 14 pulmonary medicine clinics serving 28,000 COPD patients, Sentinel Respira's exacerbation prediction engine analyzed wearable and smart inhaler data to detect impending exacerbations 24-72 hours before symptoms. In the first year, the system generated 3,400 early warning alerts, enabling outpatient intervention in 2,800 cases. COPD-related hospitalizations dropped 44%. Emergency department visits for respiratory distress decreased 52%. Average annual cost per COPD patient dropped $4,200 — driven entirely by hospitalization avoidance.

44%
Fewer hospitalizations
52%
Fewer ED visits
$4,200
Cost reduction per patient
28K
Patients monitored
Medical ICU — 32 Ventilated Beds

AI-driven ventilator weaning reducing mechanical ventilation days by 1.8

The Outcome

A 32-bed medical ICU deployed Sentinel Respira's ventilator optimization and weaning prediction engines across all mechanically ventilated patients. The weaning engine predicted extubation success with 88% accuracy, enabling safe earlier extubation in patients who had been kept ventilated unnecessarily — and preventing premature extubation in patients who would have failed. Average ventilator days dropped from 8.4 to 6.6. Failed extubation rate decreased from 18% to 10%. VAP incidence dropped 24% as a direct consequence of shorter ventilation duration.

8.4→6.6
Ventilator days reduced
18→10%
Extubation failure rate
24%
VAP reduction
88%
Weaning prediction accuracy
Children's Hospital — Pediatric Pulmonology

Asthma crisis prevention across a pediatric population

The Outcome

Sentinel Respira's asthma engine was deployed across the outpatient pediatric pulmonology practice, monitoring 4,200 children with persistent asthma through smart inhaler data and environmental trigger surveillance. The system detected exacerbation patterns an average of 24 hours before symptom onset. Emergency department visits dropped 38%. Hospitalizations for status asthmaticus dropped 52%. School absence days related to asthma decreased 41%. Parents reported that the system's proactive alerts — "your child's rescue inhaler use has increased, consider scheduling a clinic visit" — transformed their relationship with the disease from reactive to proactive.

38%
Fewer ED visits
52%
Fewer hospitalizations
41%
Fewer school absence days
4,200
Children monitored
Clinician Voices

I have managed COPD patients for twenty years, and my greatest frustration has always been the same: by the time they show up in my office or the ED, the exacerbation is already underway and the damage is being done. Sentinel Respira calls me two days before the patient calls me. That single shift — from reactive to predictive — has transformed my practice and my patients' lives.

Chief, Pulmonary Medicine
Board Certified Pulmonologist, 20 Years
Regional Medical Center

The weaning prediction engine ended the daily guessing game in my ICU. We used to round every morning, look at the RSBI, do a spontaneous breathing trial, argue about whether the patient was ready, and flip a coin. Now we have a model that integrates twenty-eight variables and tells us — with 88% accuracy — whether this patient will succeed. My extubation failure rate went from 18% to 10%. That's not statistics. That's patients who didn't get reintubated.

Medical Director, Respiratory ICU
Pulmonary & Critical Care Medicine
600-Bed Teaching Hospital

The parents told me their daughter missed forty-one fewer school days this year because of the asthma monitoring. Forty-one days. That's not a clinical metric. That's a childhood. That's what this platform does — it gives children their lives back.

Pediatric Pulmonologist
Asthma Program Director
Children's Hospital
44%
Fewer COPD hospitalizations
1.8day
Fewer ventilator days
42%
Fewer failed extubations
340+
Facilities deployed
Protect Every Breath

From the first wheeze to the last ventilator day

Schedule a clinical demonstration of Sentinel Respira — configured for your patient population, your ventilator fleet, and your pulmonary practice.

Or contact our clinical team at respira@brindwell.com