Respiratory Disease Burden and Current Assessment Paradigms
March 18th, 2026
Chronic respiratory diseases represent one of the leading causes of global morbidity and mortality, affecting over 500 million people worldwide and accounting for approximately 4 million deaths annually [1]. Chronic obstructive pulmonary disease (COPD) alone affects more than 210 million individuals globally, while asthma impacts approximately 260 million [1]. Interstitial lung diseases (ILD) and pulmonary sarcoidosis account for around 4.3 million cases, and the emerging burden of post-pandemic long COVID syndrome is estimated to affect 400 million individuals worldwide [2,3]. Beyond mortality, these conditions impose an immense economic burden-exceeding $6 trillion annually across global healthcare systems-and profoundly diminish patients' quality of life through persistent dyspnoea, exercise intolerance, and frequent exacerbations [3–6].
Objective assessment of lung function is central to the diagnosis, monitoring and management of these conditions, primarily through spirometry and related pulmonary function tests (e.g. oscillometry, diffusing capacity for carbon monoxide (DLCO)) [7]. Spirometry, which measures forced expiratory volume in one second (FEV₁), forced vital capacity (FVC), and their ratio, remains the cornerstone for detecting airflow obstruction, confirming diagnosis, staging disease severity, monitoring disease progression, and evaluating treatment response [8]. The 2019 American Thoracic Society (ATS) and European Respiratory Society (ERS) technical standards establish rigorous quality criteria that, when properly implemented, enable spirometry to serve as a reliable and reproducible biomarker of respiratory health [9].
Limitations of Contemporary Lung Function Assessment
Despite its clinical value, spirometry remains constrained by episodic and clinic-based testing, with most patients undergoing spirometry only once or twice annually [10]. This limited sampling fails to capture the dynamic burden of respiratory disease [11,12]. Additionally, spirometry is insensitive to detecting some lung pathologies (e.g., the early stages of ILD) [13–15].
The consequences of this episodic approach are substantial. allowing exacerbations—critical events that accelerate disease progression, increase mortality risk, and drive healthcare costs—to go undetected between visits [16–19]. For example, a patient with COPD may attend a clinic appointment during a period of relative stability, receiving spirometry results that fail to capture the significant lung function deterioration that occurred weeks earlier during an exacerbation, or the gradual decline that will precipitate the next acute event [20]. Similarly, treatment modifications—whether medication adjustments, pulmonary rehabilitation, or environmental interventions—cannot be optimally titrated without longitudinal data demonstrating their real-world physiological effects [21].
Moreover, access to spirometry remains inequitable. Geographic barriers, particularly in rural and underserved communities, limit the availability of trained technicians and calibrated equipment [22,23]. The COVID-19 pandemic further exposed the vulnerability of clinic-based testing models, as infection control concerns, social distancing requirements, and patient reluctance to attend healthcare facilities disrupted routine respiratory monitoring for millions of individuals precisely when it was most needed [24,25]. For patients with advanced disease, mobility limitations, or significant comorbidities, the burden of travelling to healthcare facilities for spirometry testing may itself be prohibitive [26].
Advanced laboratory tests, such as body plethysmography, DLCO, and cardiopulmonary exercise testing (CPET), offer richer insights but are impractical for widespread use due to cost, complexity and limited availability [27,28].
The Digital Health Revolution: Lessons from Cardiac Monitoring
The limitations of episodic respiratory monitoring stand in stark contrast to the transformation in cardiac care over the past two decades. Cardiac rhythm monitoring has evolved from clinic-based electrocardiography to consumer wearable devices capable of detecting cardiac events, such as atrial fibrillation, with remarkable accuracy [29,30]. Today, millions of individuals routinely track heart rate, heart rate variability, and rhythm through smartwatches, with algorithms validated against clinical standards and integrated into mainstream healthcare pathways [31]. This evolution has enabled earlier detection of cardiac abnormalities, remote monitoring of chronic conditions, and unprecedented patient engagement [32].
Respiratory medicine now stands at a similar inflection point; advances in sensor technology, mobile computing and artificial intelligence (AI) are driving the development of smartphone spirometers, wearable devices, and contactless technologies capable of delivering accurate, continuous lung function assessment at home [33–37]. While these approaches are more than five years in development, they still require rigorous validation, regulatory approval, and integration into clinical pathways before widespread adoption.
The clinical need for such technologies has never been more apparent. The long COVID pandemic has created a population of millions experiencing persistent respiratory symptoms—dyspnoea, exercise intolerance, and fluctuating lung function—that demand longitudinal monitoring to characterise recovery trajectories and guide rehabilitation [38,39]. Recent validation studies demonstrate that unsupervised home spirometry, when performed with appropriate devices and quality feedback systems, can meet ATS/ERS quality standards with >90% compliance and strong correlation (r > 0.90) to clinic-based measurements across diverse patient populations, including COPD, asthma, bronchiectasis, and ILD [40,41].
Toward Continuous, Accessible Respiratory Monitoring
Digital health technologies (DHTs) could transform respiratory care by shifting from intermittent snapshots to continuous or frequent monitoring. This approach establishes each individual's baseline and detects early deviations, signalling exacerbations or treatment failure [42,43]. Integration with other health data—such as activity, sleep, medication adherence via smart inhalers, and environmental triggers—enables holistic, personalised respiratory health management [44]. AI algorithms can synthesise these multimodal data to predict exacerbations, stratify risk, and guide clinical decision-making in ways beyond what episodic spirometry alone can achieve [45,46].
Remote monitoring models align with emerging value-based care paradigms that emphasise prevention, early intervention, and optimisation of health outcomes rather than reactive management of acute events [47,48]. By engaging patients in monitoring their respiratory health, DHTs have the potential to enhance patient engagement, health literacy, and self-management capabilities [49,50].
Objectives and Scope of This Review
Despite promising advancement, significant gaps remain in the evidence base, regulatory frameworks, and implementation pathways required to translate digital respiratory monitoring from research innovation to routine clinical practice. This narrative review critically examines the current state of digital lung function monitoring technologies, their validation against established spirometric standards, clinical applications across disease states, and the challenges that must be addressed to realise their transformative potential.
We synthesise evidence on:
Core lung function metrics and their pathophysiological significance across obstructive and restrictive respiratory diseases
Traditional laboratory-based assessment methods and their inherent limitations
Emerging digital monitoring technologies, including smartphone spirometry, wearable sensors, and AI-enhanced analysis
Clinical validation studies demonstrating accuracy, feasibility, and patient acceptability
Disease-specific applications in COPD, asthma, ILD, and long COVID syndrome
Implementation challenges, including regulatory pathways, reimbursement models, data security, and health equity considerations
Future directions for research, standardisation, and integration into connected healthcare systems
By comprehensively reviewing the evidence for digital lung function monitoring—from physiological principles to clinical validation to implementation science—this paper aims to inform clinicians, researchers, health system leaders, and policymakers about both the remarkable opportunities and the substantive challenges inherent in bringing respiratory monitoring into the digital age.
Core Lung Function Metrics
2.1 Fundamental Spirometric Parameters and Pathophysiology
Spirometry quantifies the volume of air forcibly exhaled after maximal inspiration, providing the diagnostic foundation for respiratory disease classification [51]. FVC represents the total exhaled volume of air and reflects lung size and airway patency. Normal FVC values are ≥80% of predicted based on age, height and ethnicity, with absolute values of 4.8-6.0L for adult males and 3.2-4.5L for adult females [52]. FEV₁ is a measure of the volume expelled in the first second of a forceful exhalation and reflects airway calibre, forming the basis for COPD severity grading [52]. A normal FEV₁ value is ≥80% of predicted, with absolute values of 3.5-4.5L for adult males and 2.5-3.25L for adult females [52]. The FEV₁/FVC ratio distinguishes obstructive (ratio < 0.70 or below the lower limit of normal) from restrictive patterns, though debate continues over fixed-ratio versus age-adjusted criteria [53].
Advanced metrics extend diagnostic precision: Peak expiratory flow (PEF) monitors large airway function and asthma variability. Forced expiratory flow (FEF)₂₅₋₇₅, which measures airflow in the middle portion of forced exhalation, detects early small airway dysfunction but suffers from high variability. Lung volumes measured via body plethysmography confirm true restriction (Total lung capacity (TLC) < 80% predicted) and quantify hyperinflation (elevated residual volume (RV)/TLC ratio characteristic of COPD) [54]. DLCO assesses gas exchange efficiency, with reductions distinguishing emphysema from chronic bronchitis and predicting mortality in idiopathic pulmonary fibrosis (IPF) when DLCO < 35% predicted [55]. Impulse oscillometry measures respiratory resistance and impedance during tidal breathing, detecting peripheral airway resistance (R5-R20) and reactance (X5) when spirometry remains normal—increasingly relevant in early COPD and long COVID [56].
Dynamic parameters during exercise provide functional insights: TV, respiratory rate (fR), and minute ventilation (VE) reflect the ventilatory strategy, with a shallower, quicker respiratory pattern typically seen in restrictive and obstructive lung conditions. Recovery rate—the time required for respiratory parameters to return to baseline post-exercise—has emerged as a novel biomarker of respiratory reserve, potentially predicting exacerbation risk in a manner analogous to cardiac heart rate recovery [57].
2.2 Disease-Specific Functional Signatures
COPD manifests as progressive airflow limitation (FEV₁/FVC < 0.70) with heterogeneous phenotypes:
Emphysema-predominant: Marked DLCO reduction (< 60%) and severe hyperinflation (TLC > 120%) [53].
Chronic bronchitis-predominant: Preserved DLCO with frequent exacerbations [53].
Asthma-COPD overlap: Significant bronchodilator reversibility (> 12% and > 200 mL post-bronchodilator) [53].
Functional markers add prognostic value: a six-minute walk distance (6MWD) < 350 meters predicts mortality independent of FEV₁, while oxygen desaturation (peripheral oxygen saturation (SpO₂) decline ≥ 4%) occurs in 30-50% with moderate-to-severe disease [58]. Respiratory rate increases progressively with disease severity, as defined by GOLD stages, a COPD classification based on post-bronchodilator FEV₁ (% predicted), ranging from mild (GOLD 1) to very severe (GOLD 4) airflow limitation. For example, respiratory rate rises from ~16–18 breaths/min at rest to ~22–26 at 6 minutes in GOLD 1, compared with ~22–24 to ~34–38 breaths/min in GOLD 4 [59].
ILD produces restrictive physiology with reduced TLC (< 80%), markedly reduced DLCO (often < 60% in IPF) and preserved/elevated FEV₁/FVC ratio (> 0.70). Exertional desaturation is also common despite resting normoxemia [55]. Serial FVC monitoring guides therapy; a ≥10% decline over 6-12 months indicates progression requiring treatment escalation. Home spirometry trials demonstrate the feasibility of remote weekly FVC tracking meeting quality criteria [60].
Asthma is characterised by variable, reversible obstruction; often normal spirometry between exacerbations in mild disease; significant bronchodilator reversibility; diurnal PEF variability > 20% when uncontrolled; and exercise-induced bronchoconstriction (FEV₁ decline > 10% post-exercise) in 40-90% of patients. Small airway dysfunction is becoming increasingly recognised via reduced FEF₂₅₋₇₅, and oscillometry, even with preserved FEV₁ [52].
Long COVID presents heterogeneous phenotypes, which are challenging to characterise traditionally [61]:
Pulmonary-dominant: Reduced DLCO (30-40% of cases), restrictive patterns, exertional desaturation (≥ 4% in ~25%), and small airway dysfunction on oscillometry
Dysfunctional breathing: Erratic breathing patterns, irregular TV, and hyperventilation during exercise despite preserved spirometry
Deconditioning: Reduced 6MWD (median ~434 meters, 83% predicted) with preserved lung function, suggesting cardiovascular/muscular limitation
Post-exertional malaise: Delayed symptom exacerbation 24-48 hours post-exertion requiring longitudinal monitoring
During 6MWT, long COVID patients demonstrate disproportionate VE increases, reduced breathing reserve (< 30% in ~62% of patients), and dynamic hyperinflation (Inspiratory capacity decrease > 100 mL in > 60% of patients), with prolonged recovery rates distinguishing phenotypes [61].
2.3 Environmental and Lifestyle Influences
Air pollution acutely reduces FEV₁ (20-40 mL decline per 10 μg/m³ PM₂.₅ increase), precipitating exacerbations in susceptible individuals. Longitudinal monitoring enables real-time correlation with air quality, supporting activity modification. Exercise training improves 6MWD (mean increase 44 meters) and quality of life without necessarily improving spirometry, through enhanced respiratory muscle strength, reduced hyperinflation, and improved fitness [62,63]. Smoking cessation slows FEV₁ decline from ~60 mL/year to ~30 mL/year, approaching normal age-related decline. Diurnal variation (3-8% in FEV₁, lowest morning) and seasonal patterns (50-100 mL winter reduction in COPD) must be distinguished from true progression, highlighting the value of longitudinal digital monitoring [52,53].
Assessment Methods
3. Traditional Laboratory Assessment: Methods and Limitations
Office spirometry represents the most accessible lung function assessment. Modern electronic spirometers provide real-time feedback and interpretation when performed by trained personnel using ATS/ERS-compliant equipment [51]. Technical requirements are stringent: daily 3-litre syringe calibration (accuracy ± 3%), proper coaching for maximal effort, ≥3 acceptable manoeuvres with FEV₁/FVC reproducibility (within 150 mL or within 5%), back-extrapolated volume < 5% FVC, and expiratory time ≥ 6 seconds. Despite standardisation, real-world quality is poor; 30-50% of primary care tests fail ATS/ERS criteria due to inadequate coaching, premature termination, equipment malfunction, and incorrect technique, resulting in misdiagnosis and inappropriate treatment [51].
Key limitations include patient burden (travel, time off work, physical demands), infection control concerns (60-80% utilisation drop during the COVID-19 pandemic), and most significantly, limited frequency—most patients undergo testing only 1-2 times annually (capturing 0.3% of the year) [60]. Such sparse sampling cannot reliably detect exacerbations, subtle progressive decline (30-40 mL/year may not exceed test variability), optimise therapy, identify environmental triggers, or distinguish true disease variability from measurement variation.
Body plethysmography and gas dilution measure absolute lung volumes (TLC, RV, functional residual capacity) via Boyle's Law or inert gas equilibration. Plethysmography provides gold-standard accuracy, including trapped gas and enables accurate assessment of restriction and hyperinflation. However, it requires expensive equipment ($50,000-150,000), specialised laboratories, 20-45 minutes testing, highly trained technicians, complex manoeuvres and strict contraindications. Limited geographic availability precludes routine serial monitoring.
DLCO measurement uses a single-breath technique with carbon monoxide and helium, requiring a 10-second breath-hold. It is invaluable for differential diagnosis (emphysema vs. chronic bronchitis), detecting ILD, and predicting mortality, but suffers from high variability (~10-15%), technique sensitivity, and confounding by anaemia and smoking. Availability is also limited to hospital laboratories [55].
CPET provides a comprehensive assessment during graded exercise, with continuous monitoring of gas exchange (VO₂, VCO₂), ventilation (VE, TV, respiratory rate), and cardiovascular parameters [27]. Identifies exercise limitation mechanisms, guides prognosis, and informs rehabilitation, with peak VO₂ strongly predicting survival independent of resting function [64,65]. However, CPET requires expensive equipment ($50,000-150,000), 45–90-minute duration, highly trained personnel, significant patient burden (maximal exertion), contraindications (unstable cardiac disease), and limited reproducibility due to submaximal effort and variability. These factors confine CPET to diagnostic scenarios at multi-year intervals rather than routine monitoring.
6MWT provides a simple functional measure with continuous SpO₂, heart rate, and dyspnoea monitoring [58]. Advantages include simplicity, submaximal nature, strong prognostic associations, and rehabilitation sensitivity. Standardisation is essential: ≥30-meter length corridor (50 meters preferred), consistent encouragement, two practice tests (learning effect ~30 meters), and consistent timing. Limitations include variability from motivation and corridor length, lack of mechanistic insights, insensitivity in mild disease, and confinement to supervised clinic settings, precluding home-based serial monitoring.
Respiratory muscle strength testing (maximal inspiratory/expiratory pressures) quantifies respiratory muscle function, with maximal inspiratory pressure < 60 cm H₂O (males) or < 40 cm H₂O (females) indicating weakness contributing to dyspnoea. Requires maximal sustained efforts with significant learning effects and high variability (15-20%), limiting routine application despite value in neuromuscular disorders and inspiratory muscle training guidance [27]. Complementary assessment during exercise can be obtained from oesophageal (Pes) and gastric (Pga) pressures measured continuously using soft-tipped balloon catheters; a steeper ΔPga/ΔPes slope in COPD reflects diaphragmatic de-recruitment and increased reliance on accessory inspiratory muscles. Diaphragm fatigue can further be evaluated via phrenic nerve magnetic stimulation, where falls in potentiated and superimposed twitch forces denote impaired contractile function, and respiratory muscle electromyography provides additional insight into activation patterns.
3.1 Aggregate Limitations: The Case for Digital Transformation
Traditional assessment imposes systematic barriers to chronic disease management:
Episodic sampling captures 0.3% of the year, missing disease dynamics, exacerbations, and environmental influences. Snapshot bias inadequately represents variable conditions like asthma with hourly-to-daily fluctuation. The reactive paradigm detects established decline rather than enabling early intervention. Patient burden disproportionately affects those with advanced disease requiring the most intensive monitoring. Geographic inequity concentrates advanced testing in urban tertiary centres. Resource intensity limits scalability to meet the global chronic respiratory disease burden [39].
These systematic limitations—not measurement science deficiencies—create the imperative for digital alternatives that complement and extend rather than replace traditional methods, transforming respiratory monitoring from episodic clinic-based assessment to continuous, accessible, patient-centred surveillance.
Section 4: Emerging Digital Lung Function Technologies
4.1 Technology Landscape and Classification
DHTs for respiratory monitoring encompass diverse approaches leveraging advances in sensor miniaturisation, mobile computing, computer vision, signal processing, and AI [42,43]. Smartphone-based spirometers connect handheld turbine or ultrasonic sensors to mobile applications, providing immediate FEV₁/FVC calculations with real-time quality feedback and cloud-based data storage [41]. Camera-based respiratory monitoring employs computer vision algorithms to detect chest wall motion from standard smartphone cameras, quantifying respiratory rate, tidal volume, and breathing patterns without physical contact or specialised hardware [36]. Wearable biosensors—including strain gauges, impedance plethysmography belts, and acoustic sensors—enable continuous respiratory monitoring during daily activities and sleep [66]. Smart inhalers with embedded sensors track medication adherence, inhaler technique, and environmental triggers, providing actionable feedback for asthma and COPD management [67].
AI integration transforms raw physiological data into clinical insights: Machine learning (ML) algorithms classify spirometry patterns, predict exacerbations from multimodal data streams, and identify disease phenotypes with accuracy exceeding traditional diagnostic approaches [33]. Deep learning (DL) networks analyse flow-volume curves, detect quality issues, and stratify risk using complex feature extraction impossible through manual interpretation [68]. The convergence of accessible hardware, sophisticated algorithms, and ubiquitous connectivity creates the technical foundation for medical-grade home respiratory monitoring.
4.2 Validation Studies: Accuracy and Clinical Performance
Recent validation studies demonstrate that unsupervised home spirometry can achieve accuracy comparable to clinic-based testing when appropriate devices and quality assurance systems are employed. A 2024 multi-disease cohort study (n=93) across asthma, COPD, bronchiectasis, and ILD compared ultrasonic home spirometers against laboratory-standard desktop spirometry, revealing strong correlations for FEV₁ (r=0.94) and FVC (r=0.96) with mean differences of 30 mL and 40 mL, respectively—well within clinically acceptable limits [41]. Critically, 90% of home spirometry tests met ATS/ERS quality criteria despite being unsupervised, with subjects performing weekly measurements over 12 weeks demonstrating sustained compliance (median 87% test completion rate) [41].
The INMARK trial enrolled 346 IPF patients performing weekly home spirometry over 52 weeks, establishing the feasibility of remote FVC monitoring in advanced fibrotic disease [34]. Home-measured FVC showed excellent correlation with clinic spirometry (r=0.88-0.92 across timepoints), with 82% of home tests meeting acceptability criteria [34]. Importantly, FVC decline rates measured at home closely paralleled clinic-based measurements, validating home spirometry's utility for tracking disease progression and informing treatment decisions [34]. Patients reported high satisfaction (mean score 4.2/5), with 89% preferring home testing to clinic visits for routine monitoring [34].
A qualitative co-design study of ILD patients using home spirometry revealed key insights into patient perspectives: convenience and reduced travel burden emerged as primary benefits, with patients describing testing as "simple" and device cleaning as "non-intrusive" [26]. Psychological reassurance from self-monitoring capability—particularly during periods of symptom concern between clinic appointments—represented an unexpected but valued outcome [26]. Barriers included initial technical anxiety (resolved with remote training), occasional connectivity issues, and concerns about accuracy without technician supervision (mitigated by real-time quality feedback) [26].
Systematic reviews of home spirometry across respiratory diseases confirm these individual study findings. Meta-analyses demonstrate that home spirometry shows minimal systematic bias compared to clinic measurements, with mean differences of -20 to +30 mL for FEV₁ and -10 to +40 mL for FVC, magnitudes within test-retest variability of supervised spirometry [69]. However, important caveats emerge: accuracy deteriorates in severe obstruction (FEV₁ < 30% predicted), where prolonged expiratory times and fatigue compromise technique; elderly patients (> 75 years) require more intensive training; and device-specific performance varies considerably, emphasising the importance of using validated systems meeting regulatory standards.
4.3 Clinical Applications Across Disease States
4.3.1 Chronic Obstructive Pulmonary Disease (COPD)
Digital monitoring in COPD addresses the critical challenge of early exacerbation detection—events that accelerate disease progression, increase mortality and drive most COPD-related healthcare costs [18]. Traditional paradigms rely on symptom reporting and scheduled clinic visits, often delaying recognition and treatment. Digital approaches enable the detection of physiological deterioration before symptoms become severe.
Multimodal home monitoring—integrating spirometry, symptom diaries, activity tracking, and pulse oximetry—demonstrates promise for exacerbation prediction. A 2025 BMC Digital Health study enrolled 100 COPD patients in an 18-month intervention combining smartwatch monitoring, weekly spirometry, symptom surveys, and tele-pulmonary rehabilitation [70]. Feasibility was high (96% retention) with significant reductions in COPD-related patient distress (p<0.01), although the study was powered for feasibility rather than clinical outcomes [70].
ML algorithms further enhance predictive capability. A systematic review of AI identified 41 studies employing ML/DL for COPD diagnosis, prognosis, and exacerbation prediction [68]. Support vector machines and boosting algorithms dominated ML approaches (34/41 studies) with diagnostic accuracy >78%, while DL methods (16/41 studies) achieved >72% accuracy [68]. Multimodal approaches combining spirometry, symptoms, activity, and environmental data (air quality, weather) outperformed single-modality predictions [68]. Approximately half of the studies comparing ML to DL found DL superior, suggesting that with larger datasets, DL may become the preferred approach [68].
4.3.2 Interstitial Lung Disease (ILD)
Serial FVC monitoring is central to ILD management, guiding therapeutic escalation and transplant referral timing [34]. Home spirometry transforms monitoring from episodic clinic measurements (typically 3-4 times yearly) to weekly or biweekly assessments, enabling earlier detection of progression and more granular evaluation of treatment responses.
Clinical utility extends beyond simple FVC measurement. Frequent home measurements reveal day-to-day variability missed by clinic testing: stable disease shows fluctuations of 100-150 mL, whereas active progression manifests as sustained downward trends superimposed on normal variability [34]. This distinction—trend versus noise—requires sufficient sampling frequency. Patients also report that home monitoring provides reassurance during symptom fluctuations, reducing anxiety and unnecessary clinic contacts when FVC remains stable despite subjective symptom worsening.
Implementation challenges include maintaining engagement over years (ILD progression occurs slowly), sustaining motivation during stability, and preventing overinterpretation of minor fluctuations. Successful programs combine automated anomaly detection algorithms (flagging concerning trends for clinician review) with patient education, emphasising long-term trends over individual measurements.
4.3.3 Asthma Management and Smart Inhalers
Asthma's variability makes it ideally suited for continuous digital monitoring. Traditional peak flow monitoring has been employed for decades, but suffers from poor adherence and limited integration [71]. Modern approaches combine smart inhalers tracking medication use with environmental monitoring and symptom logging.
Smart inhalers embed sensors that record each actuation, GPS location, and timestamp, enabling unprecedented insights into medication patterns, adherence, trigger identification, and exacerbation prediction [72,73]. Studies demonstrate that patients using smart inhalers show improved adherence, and fewer exacerbations and hospitalisations [74,75]. Crucially, location-based data reveals environmental triggers: patients may notice increased rescue inhaler use correlating with specific locations (workplace, specific neighbourhoods), enabling targeted environmental interventions such as air quality monitoring and HEPA filter installation.
Integration of smart inhaler data with personal air quality monitors, pollen forecasts, and weather patterns creates individualised trigger profiles. ML algorithms identify patterns invisible to clinicians or patients: subtle increases in rescue inhaler use 5 days before symptomatic exacerbations, weekend-weekday differences suggesting occupational triggers, or seasonal patterns informing preventive therapy timing [76].
4.3.4 Long COVID Phenotyping and Rehabilitation
Long COVID's heterogeneity challenges traditional assessment, with patients often experiencing persistent dyspnoea despite normal resting spirometry, fluctuating symptoms, and post-exertional malaise [61,77]. Digital monitoring provides tools for phenotype characterisation and rehabilitation guidance beyond the capabilities of clinic-based assessment.
A 2025 systematic review and meta-analysis of pulmonary rehabilitation in long COVID (23 randomised controlled trials, n=1,847 patients) demonstrated significant improvements in 6MWD (mean difference +48.5 meters, 95% CI 34.2-62.8), FEV₁ (+4.2%, 95% CI 2.1-6.3), FVC (+5.1%, 95% CI 2.8-7.4), and quality of life scores [77]. However, response heterogeneity was substantial, with some patients improving markedly, while others experienced minimal benefit or symptom exacerbation. This variability suggests distinct pathophysiological mechanisms requiring tailored interventions.
Digital monitoring enables longitudinal phenotyping by tracking respiratory rate patterns, recovery kinetics after standardised activities, and symptom correlation. Dysfunctional breathing patterns (erratic tidal volumes, inappropriate hyperventilation) can be identified for targeted breathing retraining, while patients with true exercise intolerance and prolonged recovery benefit from graded exercise programs. For those recovering from prolonged mechanical ventilation, inspiratory muscle training can help restore strength to deconditioned respiratory muscles, reducing breathlessness and improving tolerance for whole-body exercise [78,79]. Those with post-exertional malaise require pacing strategies and activity modification [80,81]. Home spirometry identifies the subset with peripheral airway dysfunction (reduced FEF₂₅₋₇₅, abnormal oscillometry) despite normal FEV₁/FVC who may benefit from inhaled therapies.
4.4 Integration with Digital Health Ecosystems
The true potential of respiratory monitoring emerges through integration with comprehensive digital health platforms. Apple Health, Google Fit, and similar ecosystems already track activity, sleep, heart rate, and electrocardiography; adding respiratory metrics creates holistic health profiles [42]. For example, a COPD patient's declining step count, increased resting heart rate, and reduced sleep quality, combined with subtle FEV₁ decline, provide earlier and more specific exacerbation warning than any single parameter.
Telemedicine platform integration enables asynchronous clinician review of home spirometry with automated flagging of concerning trends. The COVID-19 pandemic accelerated the adoption of remote consultation models; respiratory monitoring represents a natural extension [59]. Patients perform weekly home spirometry; algorithms identify significant changes (FEV₁ decline > 10%, FVC decline > 8%, or sustained downward trends); flagged results trigger clinician review and proactive outreach, preventing emergency department visits and hospitalisations.
Electronic health record (EHR) integration remains challenging but essential for clinical adoption. Siloed data in proprietary apps provides patient value but limited clinical utility. Bidirectional EHR integration—automatically importing home spirometry into medical records and providing patients with clinic-measured lung function—creates unified longitudinal views. FHIR (Fast Healthcare Interoperability Resources) standards enable data exchange, though implementation barriers, including vendor cooperation, privacy protection, and clinical workflow integration, persist [42].
AI-driven decision support synthesises multimodal data into actionable insights. Rather than presenting clinicians with overwhelming data streams (weekly spirometry, daily symptoms, continuous activity tracking), AI algorithms identify patterns, predict outcomes, and recommend interventions. A COPD patient experiencing progressive FEV₁ decline, increased rescue inhaler use, reduced activity, and worsening air quality exposure receives automated recommendations: consider a corticosteroid burst, intensify bronchodilator therapy, advise reduced outdoor activity, and schedule a clinic evaluation. These systems will augment clinical judgment and act as intelligent triage tools; however, clinicians will still need to review the recommendations and make final decisions, and therefore still require their time.
4.5 Evidence-Based and Clinical Outcomes
Evidence for digital respiratory monitoring is growing, but randomised controlled trials demonstrating improved clinical outcomes remain limited compared to feasibility and accuracy studies [59,69]. The 2025 ATS Research Statement concluded that while feasibility and accuracy are well-established, definitive evidence for mortality reduction, fewer hospitalisations, or improved quality of life requires larger, longer-duration randomised controlled trials with appropriate controls [69].
Key trials informing current understanding:
The TeleCOPD trial randomised 281 COPD patients to integrated telemonitoring (daily symptoms, spirometry, pulse oximetry with automated clinical alerts) versus usual care. No differences in hospitalisations, exacerbation frequency, or mortality were observed over 9 months, although post-hoc analysis showed fewer hospitalisations among highly adherent patients (>80%), underscoring the need for behavioural engagement and responsive clinical protocols [49].
INMARK (IPF) demonstrated that weekly home spirometry enabled earlier detection of FVC decline compared to 3-month clinic intervals, facilitating timely antifibrotic therapy initiation [34]. While not powered for hard outcomes (i.e. mortality, hospitalisation or exacerbation rates), the study established that home monitoring changes clinical management timing and decisions.
A systematic review of wearable technologies for COPD exacerbation detection (15 studies, n=1,456 patients) found that multimodal approaches (respiratory rate, activity, heart rate, symptoms) predicted exacerbations with 70-85% sensitivity and 60-75% specificity when optimised for 3–7-day prediction windows [66]. Single-parameter monitoring (respiratory rate alone, activity alone) performed poorly, emphasising the importance of integrated approaches.
4.6 Implementation Challenges and Barriers
Despite technical feasibility and clinical promise, digital respiratory monitoring faces substantial implementation barriers [42,59]:
Regulatory pathways vary across jurisdictions, with unclear requirements for software-as-medical-device classification, particularly for systems integrating multiple data sources and AI algorithms. FDA guidance on DHTs provides frameworks for clinical investigations, but pathways for multi-device, interoperable platforms remain uncertain. EU regulations similarly struggle with connected systems processing inputs from multiple sources.
Reimbursement models lag technological capabilities. Most healthcare systems reimburse clinic-based spirometry but lack codes for home monitoring interpretation, algorithm development, or remote patient management. Value-based care models theoretically incentivise preventive monitoring, but fee-for-service remains dominant, creating financial disincentives for digital adoption.
Health equity concerns risk exacerbating disparities. Digital monitoring requires smartphones, internet connectivity, technical literacy, and engagement capacity—resources unequally distributed across socioeconomic strata. Rural populations lacking broadband access, elderly individuals with limited digital literacy, and economically disadvantaged patients unable to afford devices face exclusion. Implementation must proactively address equity through device provision programs, multilingual interfaces, low-bandwidth solutions, and community health worker support.
Clinical workflow integration determines adoption success. Systems requiring manual data review without clinical decision support increase clinician burden rather than improving efficiency. Successful implementation requires automated triage, integration with existing EHR workflows, clear action protocols, and appropriate staffing models for remote monitoring program management.
Data security and privacy concerns intensify with continuous physiological monitoring. Respiratory data reveals not only health status but activity patterns, location (through environmental sensors), and potentially sensitive behavioural information. HIPAA compliance, GDPR requirements, and patient trust necessitate robust encryption, transparent data governance, and patient control over data sharing.
4.7 Future Directions and Research Priorities
The digital respiratory monitoring field requires coordinated research across multiple domains[42,59,69]:
Clinical outcomes trials must definitively demonstrate whether digital monitoring improves mortality, reduces hospitalisations, enhances quality of life, and proves cost-effective compared to usual care. Trials should be pragmatic, recruiting diverse populations, employing intention-to-treat analyses accounting for non-adherence, and following patients for sufficient duration (≥2 years) to detect hard outcomes.
Standardisation efforts should establish technical standards for device accuracy, data formats for interoperability, quality metrics for acceptable home tests, and clinical action thresholds for abnormal findings. Professional societies (ATS, ERS) should develop consensus statements analogous to spirometry standardisation guidelines.
Algorithm validation requires transparent reporting of AI/ML model development, external validation in diverse populations, assessment of bias and fairness across demographic groups, and prospective evaluation in clinical settings. Regulatory frameworks should mandate algorithm cards documenting training data, performance characteristics, limitations, and intended use.
Implementation science must elucidate effective strategies for technology adoption, patient engagement, clinician training, workflow redesign, and equitable access. Comparative effectiveness research should identify which patients benefit most from digital monitoring, optimal monitoring intensity, and cost-effective implementation models.
The vision for respiratory medicine's digital future encompasses continuous, accessible monitoring integrated with comprehensive health ecosystems, enabling early intervention, personalised management, and patient empowerment while maintaining clinical rigour, equity, and privacy. Achieving this vision requires sustained collaboration among technologists, clinicians, patients, regulators, payers, and researchers to translate technical capability into improved health outcomes across populations.
