Developing a reference of normal lung sounds in healthy Peruvian children

Artificial Intelligence Models for Pediatric Lung Sound Analysis: Systematic Review and Meta-Analysis

BACKGROUND

Pediatric respiratory diseases, including asthma and pneumonia, are major causes of morbidity and mortality in children. Auscultation of lung sounds is a key diagnostic tool but is prone to subjective variability. The integration of artificial intelligence (AI) and machine learning (ML) with electronic stethoscopes offers a promising approach for automated and objective lung sound.

OBJECTIVE

This systematic review and meta-analysis assess the performance of ML models in pediatric lung sound analysis. The study evaluates the methodologies, model performance, and database characteristics while identifying limitations and future directions for clinical implementation.

METHODS

A systematic search was conducted in Medline via PubMed, Embase, Web of Science, OVID, and IEEE Xplore for studies published between January 1, 1990, and December 16, 2024. Inclusion criteria are as follows: studies developing ML models for pediatric lung sound classification with a defined database, physician-labeled reference standard, and reported performance metrics. Exclusion criteria are as follows: studies focusing on adults, cardiac auscultation, validation of existing models, or lacking performance metrics. Risk of bias was assessed using a modified Quality Assessment of Diagnostic Accuracy Studies (version 2) framework. Data were extracted on study design, dataset, ML methods, feature extraction, and classification tasks. Bivariate meta-analysis was performed for binary classification tasks, including wheezing and abnormal lung sound detection.

RESULTS

A total of 41 studies met the inclusion criteria. The most common classification task was binary detection of abnormal lung sounds, particularly wheezing. Pooled sensitivity and specificity for wheeze detection were 0.902 (95% CI 0.726-0.970) and 0.955 (95% CI 0.762-0.993), respectively. For abnormal lung sound detection, pooled sensitivity was 0.907 (95% CI 0.816-0.956) and specificity 0.877 (95% CI 0.813-0.921). The most frequently used feature extraction methods were Mel-spectrogram, Mel-frequency cepstral coefficients, and short-time Fourier transform. Convolutional neural networks were the predominant ML model, often combined with recurrent neural networks or residual network architectures. However, high heterogeneity in dataset size, annotation methods, and evaluation criteria were observed. Most studies relied on small, single-center datasets, limiting generalizability.

CONCLUSIONS

ML models show high accuracy in pediatric lung sound analysis, but face limitations due to dataset heterogeneity, lack of standard guidelines, and limited external validation. Future research should focus on standardized protocols and the development of large-scale, multicenter datasets to improve model robustness and clinical implementation.

Deep learning-based lung sound analysis for intelligent stethoscope

Auscultation is crucial for the diagnosis of respiratory system diseases. However, traditional stethoscopes have inherent limitations, such as inter-listener variability and subjectivity, and they cannot record respiratory sounds for offline/retrospective diagnosis or remote prescriptions in telemedicine. The emergence of digital stethoscopes has overcome these limitations by allowing physicians to store and share respiratory sounds for consultation and education. On this basis, machine learning, particularly deep learning, enables the fully-automatic analysis of lung sounds that may pave the way for intelligent stethoscopes. This review thus aims to provide a comprehensive overview of deep learning algorithms used for lung sound analysis to emphasize the significance of artificial intelligence (AI) in this field. We focus on each component of deep learning-based lung sound analysis systems, including the task categories, public datasets, denoising methods, and, most importantly, existing deep learning methods, i.e., the state-of-the-art approaches to convert lung sounds into two-dimensional (2D) spectrograms and use convolutional neural networks for the end-to-end recognition of respiratory diseases or abnormal lung sounds. Additionally, this review highlights current challenges in this field, including the variety of devices, noise sensitivity, and poor interpretability of deep models. To address the poor reproducibility and variety of deep learning in this field, this review also provides a scalable and flexible open-source framework that aims to standardize the algorithmic workflow and provide a solid basis for replication and future extension: https://github.com/contactless-healthcare/Deep-Learning-for-Lung-Sound-Analysis .

Design and Comparative Performance of a Robust Lung Auscultation System for Noisy Clinical Settings

Chest auscultation is a widely used clinical tool for respiratory disease detection. The stethoscope has undergone a number of transformative enhancements since its invention, including the introduction of electronic systems in the last two decades. Nevertheless, stethoscopes remain riddled with a number of issues that limit their signal quality and diagnostic capability, rendering both traditional and electronic stethoscopes unusable in noisy or non-traditional environments (e.g., emergency rooms, rural clinics, ambulatory vehicles). This work outlines the design and validation of an advanced electronic stethoscope that dramatically reduces external noise contamination through hardware redesign and real-time, dynamic signal processing. The proposed system takes advantage of an acoustic sensor array, an external facing microphone, and on-board processing to perform adaptive noise suppression. The proposed system is objectively compared to six commercially-available acoustic and electronic devices in varying levels of simulated noisy clinical settings and quantified using two metrics that reflect perceptual audibility and statistical similarity, normalized covariance measure (NCM) and magnitude squared coherence (MSC). The analyses highlight the major limitations of current stethoscopes and the significant improvements the proposed system makes in challenging settings by minimizing both distortion of lung sounds and contamination by ambient noise.

Neonatal Heart and Lung Sound Quality Assessment for Robust Heart and Breathing Rate Estimation for Telehealth Applications

With advances in digital stethoscopes, internet of things, signal processing and machine learning, chest sounds can be easily collected and transmitted to the cloud for remote monitoring and diagnosis. However, low quality of recordings complicates remote monitoring and diagnosis, particularly for neonatal care. This paper proposes a new method to objectively and automatically assess the signal quality to improve the accuracy and reliability of heart rate (HR) and breathing rate (BR) estimation from noisy neonatal chest sounds. A total of 88 10-second long chest sounds were taken from 76 preterm and full-term babies. Six annotators independently assessed the signal quality, number of detectable beats, and breathing periods from these recordings. For quality classification, 187 and 182 features were extracted from heart and lung sounds, respectively. After feature selection, class balancing, and hyperparameter optimization, a dynamic binary classification model was trained. Then HR and BR were automatically estimated from the chest sound and several approaches were compared.The results of subject-wise leave-one-out cross-validation, showed that the model distinguished high and low quality recordings in the test set with 96% specificity, 81% sensitivity and 93% accuracy for heart sounds, and 86% specificity, 69% sensitivity and 82% accuracy for lung sounds. The HR and BR estimated from high quality sounds resulted in significantly less median absolute error (4 bpm and 12 bpm difference, respectively) compared to those from low quality sounds. The methods presented in this work, facilitates automated neonatal chest sound auscultation for future telehealth applications.

Validation of Auscultation Technologies using Objective and Clinical Comparisons

Technology is rapidly changing the health care industry. As new systems and devices are developed, validating their effectiveness in practice is not trivial, yet it is essential for assessing their technical and clinical capabilities. Digital auscultations are new technologies that are changing the landscape of diagnosis of lung and heart sounds and revamping the centuries old original design of the stethoscope. Here, we propose a methodology to validate a newly developed digital stethoscope, and compare its effectiveness against a market-accepted device, using a combination of signal properties and clinical assessments. Data from 100 pediatric patients is collected using both devices side by side in two clinical sites. Using the proposed methodology, we objectively compare the technical performance of the two devices, and identify clinical situations where performance of the two devices differs. The proposed methodology offers a general approach to verify a new digital auscultation device as clinically-viable; while highlighting the important consideration for clinical conditions in performing these evaluations.

Assessment of breath sounds at birth using digital stethoscope technology

Newborn transition is a phase of complex change involving lung fluid clearance and lung aeration. We aimed to use a digital stethoscope (DS) to assess the change in breath sound characteristics over the first 2 h of life and its relationship to mode of delivery. A commercially available DS was used to record breath sounds of term newborns at 1-min and 2-h post-delivery via normal vaginal delivery (NVD) or elective caesarean section (CS). Sound analysis was conducted, and two comparisons were carried out: change in frequency profiles over 2 h, and effect of delivery mode. There was a significant drop in the frequency profile of breath sounds from 1 min to 2 h with mean (SD) frequency decreasing from 333.74 (35.42) to 302.71 (47.19) Hz, p < 0.001, and proportion of power (SD) in the lowest frequency band increasing from 0.27 (0.11) to 0.37 (0.15), p < 0.001. At 1 min, NVD infants had slightly higher frequency than CS but no difference at 2 h.Conclusion: We were able to use DS technology in the transitioning infant to depict significant changes to breath sound characteristics over the first 2 h of life, reflecting the process of lung aeration.What is Known:• Lung fluid clearance and lung aeration are critical processes that facilitate respiration and mode of delivery can impact this• Digital stethoscopes offer enhanced auscultation and have been used in the paediatric population for the assessment of pulmonary and cardiac soundsWhat is New:• This is the first study to use digital stethoscope technology to assess breath sounds at birth• We describe a change in breath sound characteristics over the first 2 h of life and suggest a predictive utility of this analysis to predict the development of respiratory distress in newborns prior to the onset of symptoms.

Acoustic analysis of neonatal breath sounds using digital stethoscope technology

BACKGROUND

There is no published literature regarding the use of the digital stethoscope (DS) and computerized breath sound analysis in neonates, despite neonates experiencing a high burden of respiratory disease. We aimed to determine if the DS could be used to study breath sounds of term and preterm neonates without respiratory disease, and detect a difference in acoustic characteristics between them.

METHODS

A commercially available DS was used to record breath sounds of term and preterm neonates not receiving respiratory support between 24 and 48 hours after birth. Recordings were extracted, filtered, and computer analysis performed to obtain power spectra and mel frequency cepstral coefficient (MFCC) profiles.

RESULTS

Recordings from 26 term and 26 preterm infants were obtained. The preterm cohort had an average gestational age (median and interquartile range) of 32 (31-33) weeks and term 39 (38-39) weeks. Birth weight (mean and SD) was 1767 (411) g for the preterm and 3456 (442) g for the term cohort. Power spectra demonstrated the greatest power in the low-frequency range of 100 to 250 Hz for both groups. There were significant differences (P < .05) in the average power at low (100-250 Hz), medium (250-500 Hz), high (500-1000 Hz), and very high (1000-2000 Hz) frequency bands. MFCC profiles also demonstrated significant differences between groups (P < .05).

CONCLUSION

It is feasible to use DS technology to analyze breath sounds in neonates. DS was able to determine significant differences between the acoustic characteristics of term and preterm infants breathing in room air. Further investigation of DS technology for neonatal breath sounds is warranted.

Digital stethoscopes in paediatric medicine

AIM

To explore, synthesise and discuss currently available digital stethoscopes (DS) and the evidence for their use in paediatric medicine.

METHODS

Systematic review and narrative synthesis of digital stethoscope use in paediatrics following searches of OVID Medline, Embase, Scopus, PubMed and Google Scholar databases.

RESULTS

Six digital stethoscope makes were identified to have been used in paediatric focused studies so far. A total of 25 studies of DS use in paediatrics were included. We discuss the use of digital stethoscope technology in current paediatric medicine, comment on the technical properties of the available devices, the effectiveness and limitations of this technology, and potential uses in the fields of paediatrics and neonatology, from telemedicine to computer-aided diagnostics.

CONCLUSION

Further validation and testing of available DS devices is required. Comparison studies between different types of DS would be useful in identifying strengths and flaws of each DS as well as identifying clinical situations for which each may be most appropriately suited.

Building a Prediction Model for Radiographically Confirmed Pneumonia in Peruvian Children: From Symptoms to Imaging

Background

Community-acquired pneumonia remains the leading cause of death in children worldwide, and current diagnostic guidelines in resource-poor settings are neither sensitive nor specific. We sought to determine the ability to correctly diagnose radiographically confirmed clinical pneumonia when diagnostics tools were added to clinical signs and symptoms in a cohort of children with acute respiratory illnesses in Peru.

Methods

Children < 5 years of age with an acute respiratory illness presenting to a tertiary hospital in Lima, Peru, were enrolled. The ability to predict radiographically confirmed clinical pneumonia was assessed using logistic regression under four additive scenarios: clinical signs and symptoms only, addition of lung auscultation, addition of oxyhemoglobin saturation (Spo₂), and addition of lung ultrasound.

Results

Of 832 children (mean age, 21.3 months; 59% boys), 453 (54.6%) had clinical pneumonia and 221 (26.6%) were radiographically confirmed. Children with radiographically confirmed clinical pneumonia had lower average Spo₂ than those without (95.9% vs 96.6%, respectively; P < .01). The ability to correctly identify radiographically confirmed clinical pneumonia using clinical signs and symptoms was limited (area under the curve [AUC] = 0.62; 95% CI, 0.58-0.67) with a sensitivity of 66% (95% CI, 59%-73%) and specificity of 53% (95% CI, 49%-57%). The addition of lung auscultation improved classification (AUC = 0.73; 95% CI, 0.69-0.77) with a sensitivity of 75% (95% CI, 69%-81%) and specificity of 53% (95% CI, 49%-57%) for the presence of crackles. In contrast, the addition of Spo₂ did not improve classification (AUC = 0.73; 95% CI, 0.69-0.77) with a sensitivity of 40% (95% CI, 33%-47%) and specificity of 72% (95% CI, 68%-75%) for an Spo₂ ≤ 92%. Adding consolidation on lung ultrasound was associated with the largest improvement in classification (AUC = 0.85; 95% CI, 0.82-0.89) with a sensitivity of 55% (95% CI, 48%-63%) and specificity of 95% (95% CI, 93%-97%).

Conclusions

The addition of lung ultrasound and auscultation to clinical signs and symptoms improved the ability to correctly classify radiographically confirmed clinical pneumonia. Implementation of auscultation- and ultrasound-based diagnostic tools can be considered to improve diagnostic yield of pneumonia in resource-poor settings.

Changes in the breath sound spectrum during methacholine inhalation in children with asthma

BACKGROUND AND OBJECTIVE

An effort-independent breath sound analysis is expected to be a safe and simple method for clinical assessment of changes in airway function. The effects of bronchoconstriction and bronchodilation on novel breath sound parameters in asthmatic children were investigated.

METHODS

The study population included 49 children with atopic asthma (male = 33; mean age: 10.2 years). We evaluated breath sound parameters of the highest frequency of the power spectrum (HFp), frequency limiting 50% and 99% of the power spectrum (F₅₀ and F₉₉ ) and roll-off from 600 Hz to the HFp (Slope). We also assessed new parameters obtained using the ratios of sound spectrum parameters (spectrum curve indices), such as the ratio of the third and fourth power area to the total power area (P₃ /P_T and P₄ /P_T ), the ratio of the third and fourth areas to the total area under the curve (A₃ /A_T and B₄ /A_T ) and the ratio of power and frequency at 75% of HFp and 50% of HFp (RPF₇₅ and RPF₅₀ ). This was measured before and after methacholine inhalation challenge and after β₂ agonist inhalation.

RESULTS

The parameters, F₅₀ and F₉₉ , showed no changes after methacholine inhalation. Conversely, the A₃ /A_T (12.5-10.0%, P < 0.001), B₄ /A_T (7.6-5.5%, P < 0.001), RPF₇₅ (6.7-4.0 dBm/Hz, P < 0.001) and RPF₅₀ (5.8-4.3 dBm/Hz, P < 0.001) were significantly decreased. These values returned to the original level after β₂ agonist inhalation.

CONCLUSION

Spectrum curve indices indicate bronchoconstriction and bronchodilation. These parameters may play a role in the assessment of airway narrowing in asthmatic children.

Determining the outcomes of interventions to prevent respiratory syncytial virus disease in children: what to measure?

Respiratory syncytial virus (RSV) is the most common cause of viral acute lower respiratory tract illness (LRTI) in young children, and a major cause of hospital admissions and health-care utilisation globally. Substantial efforts have been made to develop RSV vaccines and vaccine-like monoclonal antibodies to prevent acute RSV LRTI. Prevention of acute disease could improve long-term lung health, with potential effects on wheezing, asthma, and chronic lung disease. This Personal View describes assessments that should be initiated during clinical trials and continued after licensure to fully evaluate the effect of RSV preventive interventions. These assessments include recording the incidence of RSV-specific LRTI and all-cause LRTI through two RSV seasons, and assessment of the prevalence and severity of recurrent wheezing or asthma in children aged up to 6 years. Standardised assessments in diverse settings are needed to fully determine the effect of interventions for the prevention of RSV disease.

Digital stethoscopes compared to standard auscultation for detecting abnormal paediatric breath sounds

Our study aimed to objectively describe the audiological characteristics of wheeze and crackles in children by using digital stethoscope (DS) auscultation, as well as assess concordance between standard auscultation and two different DS devices in their ability to detect pathological breath sounds. Twenty children were auscultated by a paediatric consultant doctor and digitally recorded using the Littman™ 3200 Digital Electronic Stethoscope and a Clinicloud™ DS with smart device. Using spectrographic analysis, we found those with clinically described wheeze had prominent periodic waveform segments spanning expiration for a period of 0.03-1.2 s at frequencies of 100-1050 Hz, and occasionally spanning shorter inspiratory segments; paediatric crackles were brief discontinuous sounds with a distinguishing waveform. There was moderate concordance with respect to wheeze detection between digital and standard binaural stethoscopes, and 100% concordance for crackle detection. Importantly, DS devices were more sensitive than clinician auscultation in detecting wheeze in our study.

CONCLUSION

Objective definition of audio characteristics of abnormal paediatric breath sounds was achieved using DS technology. We demonstrated superiority of our DS method compared to traditional auscultation for detection of wheeze. What is Known: • The audiological characteristics of abnormal breath sounds have been well-described in adult populations but not in children. • Inter-observer agreement for detection of pathological breath sounds using standard auscultation has been shown to be poor, but the clinical value of now easily available digital stethoscopes has not been sufficiently examined. What is New: • Digital stethoscopes can objectively define the nature of pathological breath sounds such as wheeze and crackles in children. • Paediatric wheeze was better detected by digital stethoscopes than by standard auscultation performed by an expert paediatric clinician.

A clinical method for detecting bronchial reversibility using a breath sound spectrum analysis in infants

BACKGROUND

Using a breath sound analyzer, we investigated clinical parameters for detecting bronchial reversibility in infants.

METHODS

A total of 59 infants (4-39 months, mean age 7.8 months) were included. In Study 1, the intra- and inter-observer variability was measured in 23 of 59 infants. Breath sound parameters, the frequency at 99% of the maximum frequency (F₉₉), frequency at 25%, 50%, and 75% of the power spectrum (Q₂₅, Q₅₀, and Q₇₅), and highest frequency of inspiratory breath sounds (HFI), and parameters obtained using the ratio of parameters, i.e. spectrum curve indices, the ratio of the third and fourth area to total area (A₃/A_T and B₄/A_T, respectively) and ratio of power and frequency at F₇₅ and F₅₀ (RPF₇₅ and RPF₅₀), were calculated. In Study 2, the relationship between parameters of breath sounds and age and stature were studied. In Study 3, breath sounds were studied before and after β₂ agonist inhalation.

RESULTS

In Study 1, the data showed statistical intra- and inter-observer reliability in A₃/A_T (p=0.042 and 0.034, respectively) and RPF₅₀ (p=0.001 and 0.001, respectively). In Study 2, there were no significant relationships between age, height, weight, and BMI. In Study 3, A₃/A_T and RPF₅₀ significantly changed after β₂ agonist inhalation (p=0.001 and p<0.001, respectively).

CONCLUSIONS

Breath sound analysis can be performed in infants, as in older children, and the spectrum curve indices are not significantly affected by age-related factors. These sound parameters may play a role in the assessment of bronchial reversibility in infants.

Adaptive Noise Suppression of Pediatric Lung Auscultations With Real Applications to Noisy Clinical Settings in Developing Countries

GOAL

Chest auscultation constitutes a portable low-cost tool widely used for respiratory disease detection. Though it offers a powerful means of pulmonary examination, it remains riddled with a number of issues that limit its diagnostic capability. Particularly, patient agitation (especially in children), background chatter, and other environmental noises often contaminate the auscultation, hence affecting the clarity of the lung sound itself. This paper proposes an automated multiband denoising scheme for improving the quality of auscultation signals against heavy background contaminations.

METHODS

The algorithm works on a simple two-microphone setup, dynamically adapts to the background noise and suppresses contaminations while successfully preserving the lung sound content. The proposed scheme is refined to offset maximal noise suppression against maintaining the integrity of the lung signal, particularly its unknown adventitious components that provide the most informative diagnostic value during lung pathology.

RESULTS

The algorithm is applied to digital recordings obtained in the field in a busy clinic in West Africa and evaluated using objective signal fidelity measures and perceptual listening tests performed by a panel of licensed physicians. A strong preference of the enhanced sounds is revealed.

SIGNIFICANCE

The strengths and benefits of the proposed method lie in the simple automated setup and its adaptive nature, both fundamental conditions for everyday clinical applicability. It can be simply extended to a real-time implementation, and integrated with lung sound acquisition protocols.