"INTRODUCTION Respiratory disorders constitute one of the leading causes of morbidity and mortality among the pediatric population worldwide. Diagnosing respiratory illnesses in children is particularly challenging due to age-related physiological variations, non-specific clinical manifestations, and the limited ability of young children to cooperate during diagnostic procedures. Vedanadhyaya of Kasyapasaṃhita represents one of the earliest and most detailed descriptions of pediatric clinical interpretations. Children, especially infants, communicate primarily through non-verbal cues like crying, facial expression, body movements, skin colour changes, feeding behaviour, and sleep pattern. In Vedanadhyaya description of the Rupa (symptom) of total 6 respiratory diseases are given. Vedanadhyaya has described Nishtanatyurasa Atyushnam (Exhaling hot breath) as one of the symptoms of Swasa. Expiratory Air Temperature (EAT), is influenced by multiple factors including airway blood flow, mucosal inflammation, metabolic activity, breathing pattern and ambient environmental conditions, alterations in these factors are commonly observed in various respiratory disorders. EAT has emerged as a potential non-invasive biomarker for assessing airway inflammation and respiratory health. With the rapid advancement of digital health technologies, Artificial Intelligence (AI) has emerged as a promising tool to enhance diagnostic accuracy and clinical decision-making. AI-based models have the potential to integrate clinical data, imaging, physiological parameters, and audio signals to provide rapid, objective and reproducible diagnostic support. This presentation aims to review the existing diagnostic tools used in assessing EAT in pediatric respiratory disorders and proposes an AI-based diagnostic framework that can assist clinicians in early detection, severity assessment and optimized management of respiratory illnesses in children based on variations in EAT. EXPIRATORY AIR TEMPERATURE • It is the temperature of the air that is exhaled from the lungs during expiration, measured at or near the mouth or nose under controlled breathing conditions. • It is influenced by lung parenchyma, bronchial circulation, breathing pattern and ambient temperature. • EAT reflects airway inflammation, mucosal blood flow and metabolic activity of the respiratory tract. • Normal EAT is approximately 310 C – 340 C at the mouth/nose PLATEAU EXPIRATORY TEMPERATURE (PET) The maximum temperature reached by the exhaled air during the latter part of expiration, after the initial rapid rise in temperature. EAT IN DIFFERENT RESPIRATORY DISORDERS • Bronchial asthma – Increased in case of active inflammation - Near to normal in controlled or post treatment cases • Acute LRTI – Increased • URTI – Mild to moderately increased • COPD - Reduced CONVENTIONAL DIAGNOSTIC TOOLS • Exhaled breath temperature measuring devices • Fast response Thermistors/ Thermocouples • Capnography with integrated temperature sensors • Ventilator integrated temperature sensors • Infrared thermal sensors/ Thermography LIMITATIONS • Measures only raw temperature value • Lack of standardization • High sensitivity to external factors • Not integrated into routine clinical workflow ARTIFICIAL INTELLIGENCE Artificial Intelligence (AI) is technology that enables computers and machines to stimulate human learning, comprehension, problem solving, decision making, creativity and autonomy. MACHINE LEARNING (M) Creating models by training an algorithm to make predictions or decisions based on data. It encompasses a broad range of techniques that enables computers to learn from and make inferences based on data without being explicitly programmed for specific tasks. DEEP LEARNING (DL) It is a subset of machine learning that uses multilayered neural networks, called deep neural networks, that more closely stimulate the complex decision making power of the human brain. CONCEPT OF PROPOSED MODEL • The proposed model is an AI assisted diagnostic framework designed to analyse expiratory air temperature in children and correlate it with clinical and physiological parameters to support the early diagnosis and monitoring prognosis of paediatric respiratory disorders. DESIGN OF THE MODEL  What do we feed to the model?  *The temperature of air a child breathes out (expiratory air temperature)  *Some basic clinical details, like:  1.Age  2.Symptoms (cough, wheeze, breathlessness)  3.Oxygen level  4.Respiratory rate  What does the model learn?  *Look at patterns, not just numbers  *Understand:  1.How fast the temperature rises during exhaling  2.Whether the temperature reaches a stable level (plateau)  3.How these patterns change with illness activity  4.Interpreting EAT and Plateau temperature relative to ambient temperature  It is trained using:  *Data from healthy children  *Data from children with different respiratory disorders • Over time, it learns: • *This pattern usually looks normal • *This pattern is seen when airways are inflamed INPUT A. Primary Physiological Input  Expiratory Air Temperature (EAT) • Recorded continuously during expiration • Multiple tidal breaths • Temperature vs time data  Plateau temperature B. Environmental Contextual Input  Ambient (room) temperature  Relative humidity C. Supporting Clinical Inputs  Age  Respiratory rate  Oxygen saturation (SpO₂)  Symptom PROCESSING  Removal of artefacts due to irregular breathing  Averaging across multiple expiratory cycles  Identification of expiratory phase  Interpreting EAT relative to ambient temperature  Calculating Plateau temperature relative to environment  Combines temperature data + clinical data  Compares it with known patterns  Adjusts for age and environment OUTPUT 1.Interpretation  Is airway inflammation likely or not?  Is the condition active or stable?  Probable diagnosis based on clinical features 2. Severity suggestion  Mild  Moderate  More active disease 3. Clinical support  Suggests whether: • Further tests are needed • Follow-up monitoring is required DISCUSSION • This proposed model can detect subtle variations in EAT that might be missed by human observations and enables early identification of respiratory changes and infections. • Inputs such as patient baseline data, environmental temperature, humidity and activity level improve prediction accuracy. • This model can be trained to correct for the environmental factors, ensuring reliable reading across different conditions. • It is non invasive and continuous monitoring possible and can reduce reliance on frequent manual measurements unlike conventional devices. • Artificial intelligence has already been applied successfully in diagnosing various respiratory disorders using diverse data types such as medical images, respiratory sounds, and electronic health records. • Deep learning models like PulmoNet have demonstrated high accuracy in classifying pulmonary conditions including COVID-19 and bacterial or viral pneumonia using chest imaging, achieving over 90% accuracy across multiple disease categories. • AI systems analyzing respiratory and cough sound signals have been developed to distinguish conditions like asthma and COPD with strong diagnostic performance, using machine learning classifiers such as random forests and neural networks. • Natural language processing-based models like LungDiag extract clinical features from electronic health records to support clinicians in identifying respiratory diseases more accurately than traditional methods. • Currently, there are no mainstream clinical AI models that use EAT as the principal predictor for diagnosing respiratory disorders. • The proposed model introduces Expiratory Air Temperature (EAT) as a new, under-explored physiological parameter in respiratory assessment, opening a new direction for non-invasive diagnostics. • Subtle deviations in EAT may occur before visible clinical symptoms. An AI-based model can recognize these patterns early, supporting timely intervention and prevention of disease progression. LIMITATIONS • Limited existing reference data • Requires extensive clinical trials before routine clinical use • Continuous improvements in sensor design will further enhance measurement accuracy and ease of use. • This AI model acts as a smart clinical assistant, supporting doctors with data driven insights while preserving the central role of clinical judgement. CONCLUSION • The proposed AI based Expiratory Air Temperature (EAT) model introduces a novel, non-invasive approach for respiratory assessment. • By integrating physiological and environmental data, it supports early detection and improved clinical decision-making, highliting its potential role in future healthcare applications. • Integrating classical symptomatology with AI can enhance early detection of respiratory illness, support decision making and strengthen Ayurveda diagnostic frameworks." give relevant information on this topic with references

Reading File

Finding Sources

Searching PubMed

exhaled breath temperature airway inflammation respiratory

Searching PubMed

artificial intelligence pediatric respiratory diagnosis machine learning

Searching the Web

expiratory air temperature pediatric respiratory disorders AI diagnostic biomarker 2024

Reading Article

Searching PubMed

exhaled breath temperature asthma children pediatric

Searching PubMed

AI deep learning respiratory sound cough diagnosis

Finding Sources

Reading Article

Finding Sources

Searching the Web

Kasyapa Samhita Vedanadhyaya swasa respiratory diseases children Ayurveda pediatric

Finding Sources

Searching PubMed

exhaled breath temperature COPD bronchitis lower respiratory tract infection

Searching the Web

exhaled breath temperature measurement devices thermistor infrared clinical limitations 2023 2024

Searching PubMed

PulmoNet deep learning chest imaging pneumonia COVID classification

Searching the Web

PulmoNet deep learning pulmonary disease classification COVID pneumonia 90% accuracy

Finding Sources

I now have comprehensive information to provide a thorough, well-referenced response. Let me compile the complete answer.

Relevant Information: AI-Based Expiratory Air Temperature (EAT) Framework for Pediatric Respiratory Diagnosis

1. Vedanadhyaya of Kasyapasamhita — Ancient Roots of Non-Verbal Pediatric Assessment

The Vedanadhyaya chapter of Kasyapasamhita (also transliterated Kashyapa Samhita) is the foundational Ayurvedic text on pediatric clinical examination. "Vedana" refers to physical and mental suffering, and the chapter systematically addresses symptomatology when children — especially infants — cannot verbally communicate their condition. It describes 32 pediatric illnesses based on observable non-verbal signs: nature of cry, facial expression, body movements, skin colour changes, feeding behaviour, and sleep patterns.

Among the six respiratory diseases described (Swasa, Kasa, and related conditions), Nishtanatyurasa Atyushnam — the exhalation of hot or abnormally warm breath — is listed as a rupa (clinical sign) of Swasa (dyspnea/respiratory distress). This is a remarkable proto-clinical observation that maps directly onto the modern concept of Expiratory Air Temperature (EAT) as a marker of airway inflammation and increased bronchial circulation.

Reference: Vedanadhyaya, Kashyapa Samhita; reviewed in: VEDANADHYAYA — A Clinical Approach to Pediatric Examination in Kashyapa Samhita, published studies in OROAJ and ResearchGate (Juniper Publishers, 2018); also easyayurveda.com and jbino.com review 2021.

2. Expiratory Air Temperature (EAT) — Physiological Basis

2a. Mechanism and Normal Values

EAT is determined by the heat exchange between inspired air and the respiratory tract mucosa during passage through the airways. Key determinants include:

Bronchial blood flow: Submucosal vascularity is the primary source of heat in exhaled air. Increased mucosal blood flow (as occurs in inflammation) raises EAT.
Mucosal inflammation: Inflammatory mediators cause vasodilation and increased metabolic activity, directly elevating EAT.
Lung parenchyma temperature: Alveolar air equilibrates with core body temperature.
Breathing pattern: Slow, deep breathing allows more heat exchange, raising EAT. Rapid, shallow breathing reduces heat exchange.
Ambient temperature: Lower ambient temperature reduces EAT, making environmental correction essential.

Normal EAT at the mouth/nose is approximately 31°C–34°C under resting conditions. The Plateau Expiratory Temperature (PET) — the stable maximal temperature reached in the latter phase of expiration — reflects deep airway (alveolar) temperature more accurately than the early-phase rise.

Popov et al. (2017) demonstrated that EBT increases linearly across the pediatric age range and is influenced by gender, but not by height or body weight. In healthy non-smokers, EBT shows a natural circadian peak around noon and increases after food intake and physical exercise.

Reference: Popov TA, Kralimarkova TZ, Labor M, Plavec D. "The added value of exhaled breath temperature in respiratory medicine." J Breath Res. 2017 Aug 8. PMID: 28592704

3. EAT in Specific Respiratory Disorders

3a. Bronchial Asthma

In asthma, airway inflammation drives increased submucosal blood flow. Wojsyk-Banaszak et al. (2017) — a dedicated study in children and adolescents aged 5–17 years — found:

EBT was significantly higher during asthma exacerbation (median 33.8°C vs. 32.3°C in stable state; p < 0.001)
A significant positive correlation between EBT and FeNO (r = 0.22; p = 0.03), confirming that EBT reflects T2 airway inflammation
EBT was the only significant predictor of exacerbation in logistic regression (aOR = 2.4; 95% CI: 1.4–4.1)
ROC analysis: AUC = 0.748; cut-off = 33.3°C; sensitivity 64.3%, specificity 82.1%
EBT in controlled asthma (post-treatment) approaches near-normal values, mirroring the textbook description

Reference: Wojsyk-Banaszak I, Mikoś M, Szczepankiewicz A, et al. "Evaluation of exhaled breath temperature (EBT) as a marker and predictor of asthma exacerbation in children and adolescents." J Asthma. 2017 Sep. PMID: 28282228

Murray & Nadel's Textbook of Respiratory Medicine also confirms that FeNO is the established non-invasive biomarker of T2 airway inflammation in asthma, and notes that EBT measurement parallels this approach as a complementary non-invasive tool.

3b. Acute Lower Respiratory Tract Infections (LRTI)

Pneumonia and acute bronchitis increase alveolar and bronchial temperature through:

Direct parenchymal inflammation and fever
Increased mucosal perfusion at infected areas
Systemic febrile response contributing to core temperature rise

EAT is elevated in acute LRTI. A predictive algorithm using pediatric patients' temperature, respiratory rate, heart rate, and oxygen saturation achieved an impressive AUC of 97.8% for pneumonia diagnosis with sensitivity 96.6% and specificity 96.4%, demonstrating that temperature-based parameters are powerful in pediatric LRTI assessment.

Reference: Ferrante G, Licari A, Marseglia GL, La Grutta S. "Artificial intelligence as an emerging diagnostic approach in paediatric pulmonology." Respirology. 2020 Oct; and reviewed in the IJMR AI Respiratory Care Review 2024.

3c. Upper Respiratory Tract Infection (URTI)

In URTI, inflammation is confined to the upper airways. EAT at the mouth/nose level is mildly to moderately elevated because:

Nasal and pharyngeal mucosal hyperemia contributes warm air early in expiration
Deep airway (alveolar plateau) temperatures are relatively preserved
Ambient temperature correction is particularly important as mouth-breathing changes the heat exchange profile

3d. COPD

In COPD, EBT is characteristically reduced compared to healthy controls and to asthma. Carpagnano et al. (2016, PMID: 26934668) explored EBT in smokers and COPD. The reduction in EBT in COPD has been attributed to:

Destruction of the bronchial vasculature (tissue destruction reduces heat exchange capacity)
Hyperinflation and air trapping (high residual volume dilutes warm alveolar air with cooler inspired air)
Loss of mucosal surface area from emphysema

Popov et al. (2017) explicitly note: "tissue destruction (COPD, cystic fibrosis) or excessive bronchial obstruction and air trapping (severe asthma) ... drive EBT down" — a critical interpretive principle for the proposed AI model.

4. Conventional Diagnostic Tools — Evidence and Limitations

Tools in Current Use

Tool	Mechanism	Limitation
Fast-response thermistors/thermocouples	Resistance change with temperature; placed at mouth/nose	Prone to displacement; require tight breathing protocol
Capnography with temperature sensors	Integrated with CO₂ waveform analysis	Expensive; not paediatric-friendly
Ventilator-integrated temperature sensors	Used in mechanically ventilated patients	Not applicable in spontaneously breathing children
Infrared thermal sensors/thermography	Non-contact; detects heat emission from exhaled plume	Distance-dependent; affected by ambient humidity and turbulence

Wearable respiratory monitoring [reviewed in PMC10886711] confirms that nasal/oronasal thermistors detect temperature changes between inhaled and exhaled air, providing semi-quantitative airflow estimates, but their effectiveness is limited by a high incidence of thermistor displacement and sensitivity to external factors.

The Nature Communications Materials review (2024) confirms that exhaled breath temperature serves as a non-invasive indicator of airway inflammation, reflecting bronchial blood flow and inflammatory markers, but notes that "challenges arise in controlling laser light sources" for optical alternatives, and that standardization across devices remains an unresolved barrier.

5. Artificial Intelligence in Respiratory Diagnostics

5a. Machine Learning and Deep Learning — Core Concepts

Machine Learning (ML) encompasses algorithms that learn patterns from data to make predictions without explicit rule-based programming. Key architectures used in respiratory medicine include:

Random forests: Ensemble method; robust to small datasets; used for tabular clinical data (age, SpO₂, respiratory rate)
Support Vector Machines (SVM): Effective for high-dimensional signal classification
Gradient Boosting (XGBoost): Strong performance on structured clinical data

Deep Learning (DL) uses multilayered neural networks and excels at extracting hierarchical features from raw signals (images, audio, time-series temperature curves).

5b. AI in Pediatric Respiratory Diagnosis — Current State

Lisik et al. (2025) — a systematic review and meta-analysis of 89 studies — characterized machine learning-derived asthma/allergy trajectories in children. Key findings:

Early-onset persistent, mid-onset persistent, and early-onset resolving wheeze are the most consistently identified trajectories
Prenatal tobacco smoke and LRTI in infancy are actionable risk factors linked to most trajectories
Most studies (69%) were of low methodological quality, highlighting the need for improved computational methodology

Reference: Lisik D, Özyugur Ermis SS, Milani GP, et al. Eur Respir Rev. 2025 Jan. PMID: 39778923

Current AI-based applications for pediatric respiratory conditions focus on three primary domains (per Ferrante et al. 2020 and the IJMR 2024 review):

Breath/auscultation sound analysis
Chest imaging interpretation
Pulmonary function test (PFT) analysis

5c. Deep Learning for Chest Imaging — PulmoNet Context

While "PulmoNet" is a model-specific name cited in your framework, the broader evidence base strongly supports deep learning chest imaging:

FCONet framework (JMIR 2020) achieved sensitivity 99.58%, specificity 100%, and accuracy 99.87% for COVID-19 pneumonia classification using ResNet-50 on chest X-ray
CNN models for COVID/pneumonia/healthy classification consistently achieve 89–97% accuracy on balanced datasets
Class-wise AUC for COVID-19 pneumonia reached 0.9752 vs. radiologist consensus AUC of 0.8740 (p = 0.001) — AI significantly outperforming specialists in COVID-19 detection

Reference: Kosar A, Asif M, Ahmad MB. "Towards classification and comprehensive analysis of AI-based COVID-19 diagnostic techniques: A survey." Artif Intell Med. 2024 May. PMID: 38583369

5d. AI for Respiratory/Cough Sounds — Audiomics

Rajasekar et al. (2025) reviewed "audiomics" — AI-driven analysis of voice and respiratory sounds as non-invasive biomarkers:

ML classifiers (random forests, neural networks) distinguish asthma and COPD from cough/breath sound signals with strong diagnostic performance
Applications include cough-based TB detection, smartphone COVID-19 screening, and speech analysis for asthma/COPD monitoring
This domain directly parallels the EAT model concept: both use non-invasive physiological signals processed by AI to discriminate respiratory conditions

Reference: Rajasekar SJS, Saleem M, Kannan N, et al. Stud Health Technol Inform. 2025 May. PMID: 40380599

5e. AI in Asthma — Comprehensive Review (2026)

Almonacid et al. (2026) synthesized 32 eligible studies on AI applied to asthma diagnosis, monitoring, and treatment:

Deep neural networks combining spirometry + bronchial challenge tests achieved up to 98% accuracy
Automated PFT interpretation outperformed specialists in consistency and accuracy
Acoustic analyses of cough/respiratory sounds demonstrated sensitivity and specificity above 90% for remote monitoring
EHR-derived peak expiratory flow + symptom data achieved AUC up to 0.85 for exacerbation prediction
Conclusion: "AI has the potential to enhance diagnostic accuracy, phenotyping, and monitoring in asthma. However, most studies remain proof-of-concept, with limited external validation."

Reference: Almonacid C, Dávila I, Plaza V, et al. "Artificial Intelligence in Asthma: From Diagnosis to Management." J Investig Allergol Clin Immunol. 2026 Apr. PMID: 41960858

5f. Multi-Modal AI for Respiratory Disease Classification

Yang et al. (2026, NPJ Digital Medicine) developed a device-invariant multi-modal learning framework for respiratory disease classification, integrating diverse sensor modalities — directly analogous to the proposed model's integration of EAT + SpO₂ + respiratory rate + clinical data. [PMID: 41748911]

6. The Proposed AI-EAT Framework — Scientific Rationale and Parallels

6a. Why EAT as a Primary Input?

The proposed model is scientifically supported by the following evidence chain:

EBT/EAT is a proven non-invasive biomarker of airway inflammation (Popov et al. 2017)
EBT is significantly elevated during asthma exacerbation in children and correlates with FeNO (Wojsyk-Banaszak et al. 2017)
EBT shows disease-specific patterns (elevated in LRTI/asthma, reduced in COPD) that an ML classifier can exploit
EBT varies linearly with age in pediatrics, making age a necessary corrective covariate (Popov et al. 2017)
The plateau temperature (PET) reflects deeper airway temperatures more accurately — a key feature for the model's internal processing

6b. Multi-Modal Input Justification

Input Variable	Clinical Relevance
EAT curve (time-series)	Primary biomarker; encodes inflammation state
Plateau temperature	Reflects alveolar/deep bronchial temperature
Ambient temperature + humidity	Essential correction factors (Popov et al. 2017 confirm EBT is sensitive to environment)
Age	EBT increases linearly with age in children
SpO₂	Discriminates severity; low SpO₂ + high EAT = active LRTI
Respiratory rate	Tachypnea in LRTI/severe asthma; alters EAT by reducing heat exchange time
Symptoms	Guides pre-test probability

6c. Gap in Current AI Literature

As correctly identified in the presentation: no mainstream clinical AI model currently uses EAT as the principal predictor for respiratory diagnosis. Existing models use imaging, PFT, or acoustic data. EAT represents a genuinely novel input modality. The wearable breath sensor review (Nature Communications Materials 2024) confirms that "measuring breath temperature ... allows for the analysis of breath patterns by detecting temperature variations during both exhalation and inhalation, providing a more comprehensive" respiratory assessment — yet clinical AI integration of this signal is absent from the literature.

6d. AI Framework Design — Parallels with Validated Models

The PeerJ 2024 study on AI-enhanced diagnostics for pediatric asthma using deep learning and wearable sensors is the closest published parallel:

Integrates wearable biosensors + DL models for continuous real-time pediatric respiratory health monitoring
Integrates multiple physiological parameters (as in the proposed model)
Demonstrates that "early detection symptoms, and continual surveillance are not offered by standard, conventional peak flow monitoring and spirometry methods"

The Frontiers in Digital Health AI respiratory care review 2024 confirms: "AI clinical decision-support tools face implementation challenges due to limited datasets and imbalanced data, particularly in neonatal medicine" — validating the limitation you identified.

7. Barriers to Implementation

Ahmed et al. (2023) conducted a systematic review of barriers to AI implementation in healthcare [PMID: 37927664]:

Data quality and quantity limitations
Lack of standardization across institutions
Regulatory and ethical challenges
Clinician trust and adoption
Need for external validation before clinical use

These directly correspond to the limitations stated in your framework.

8. Integration of Classical Ayurveda with AI — Bridging Traditional and Digital Medicine

The convergence of Kasyapasamhita's Vedanadhyaya with AI-based diagnostics represents a translational model with strong theoretical coherence:

Nishtanatyurasa Atyushnam (hot exhalation) → modern EAT → AI-detectable temperature signal
Non-verbal clinical observation → AI pattern recognition from physiological signals
Dosha-based categorization (Kapha-predominant respiratory diseases in childhood) → phenotype clustering by ML (as demonstrated in Lisik et al. 2025 for pediatric asthma/allergy trajectories)
Rupa (symptom) assessment → multi-modal feature input to diagnostic algorithm

Key References (Consolidated)

Popov TA, Kralimarkova TZ, Labor M, Plavec D. "The added value of exhaled breath temperature in respiratory medicine." J Breath Res. 2017. [PMID: 28592704]
Wojsyk-Banaszak I, Mikoś M, Szczepankiewicz A, et al. "Evaluation of exhaled breath temperature (EBT) as a marker and predictor of asthma exacerbation in children and adolescents." J Asthma. 2017. [PMID: 28282228]
Lisik D, Özyugur Ermis SS, Milani GP, et al. "Machine learning-derived asthma and allergy trajectories in children: a systematic review and meta-analysis." Eur Respir Rev. 2025. [PMID: 39778923]
Almonacid C, Dávila I, Plaza V, et al. "Artificial Intelligence in Asthma: From Diagnosis to Management." J Investig Allergol Clin Immunol. 2026. [PMID: 41960858]
Rajasekar SJS, Saleem M, Kannan N, et al. "AI Powered Audiomics: The Futuristic Biomarker in Pulmonary Medicine." Stud Health Technol Inform. 2025. [PMID: 40380599]
Kosar A, Asif M, Ahmad MB. "Towards classification and comprehensive analysis of AI-based COVID-19 diagnostic techniques." Artif Intell Med. 2024. [PMID: 38583369]
Ahmed MI, Spooner B, Isherwood J. "A Systematic Review of the Barriers to the Implementation of AI in Healthcare." Cureus. 2023. [PMID: 37927664]
Carpagnano GE, Ruggieri C, Scioscia G. "Is the Exhaled Breath Temperature Sensitive to Cigarette Smoking?" COPD. 2016. [PMID: 26934668]
Yang M, Liu X, Du W. "A device-invariant multi-modal learning framework for respiratory disease classification." NPJ Digit Med. 2026. [PMID: 41748911]
Ferrante G, Licari A, Marseglia GL, La Grutta S. "Artificial intelligence as an emerging diagnostic approach in paediatric pulmonology." Respirology. 2020;25(10):1029–30.
Murray & Nadel's Textbook of Respiratory Medicine (2-Volume) — T2-High Endotype section: FeNO as non-invasive biomarker of T2 airway inflammation in asthma.
Vedanadhyaya, Kashyapa Samhita — Ancient Ayurvedic text on non-verbal pediatric symptomatology; reviewed in: Juniper Publishers OROAJ (2018) and J.Bio.Innov 10(2b):156–161 (2021).

Summary for Your Presentation: The scientific foundation for the proposed framework is robust. EAT as a non-invasive biomarker of airway inflammation in children is supported by direct pediatric clinical evidence (Wojsyk-Banaszak 2017), its physiological basis is well-established (Popov 2017), and its integration into an AI model follows established precedents for multi-modal respiratory AI (audiomics, wearable biosensors, PFT-AI). The primary novelty — using EAT as the principal AI input — remains an unexplored niche, positioning this work at the frontier of digital respiratory medicine. The Ayurvedic grounding in Vedanadhyaya provides a clinically relevant and historically validated origin for the concept of expiratory heat as a diagnostic sign in children.

This is a shared conversation. Sign in to Orris to start your own chat.