Automated Interpretation of Lung Sounds by Deep Learning in Children With Asthma: Scoping Review and Strengths, Weaknesses, Opportunities, and Threats Analysis (Preprint) DOI
Isabelle Ruchonnet‐Metrailler, Johan N Siebert, Mary-Anne Hartley

et al.

Published: Oct. 14, 2023

BACKGROUND The interpretation of lung sounds plays a crucial role in the appropriate diagnosis and management pediatric asthma. Applying artificial intelligence (AI) to this task has potential better standardize assessment may even improve its predictive potential. OBJECTIVE This study aims objectively review literature on AI-assisted auscultation for asthma provide balanced strengths, weaknesses, opportunities, threats. METHODS A scoping sound analysis children with was conducted across 4 major scientific databases (PubMed, MEDLINE Ovid, Embase, Web Science), supplemented by gray search Google Scholar, identify relevant studies published from January 1, 2000, until May 23, 2023. strategy incorporated combination keywords related AI, pulmonary auscultation, children, quality eligible assessed using ChAMAI (Checklist Assessment Medical Artificial Intelligence). RESULTS identified 7 out 82 (9%) be included through an academic search, while 11 250 (4.4%) were considered but not subsequent assessment. All had poor medium scores, mostly due absence external validation. Identified strengths improved accuracy AI allow prompt early diagnosis, personalized strategies, remote monitoring capabilities. Weaknesses heterogeneity between lack standardization data collection interpretation. Opportunities coordinated surveillance, growing sets, new ways collaboratively learning distributed data. Threats both generic field medical (loss interpretability) also specific use case, as clinicians might lose skill auscultation. CONCLUSIONS To achieve opportunities automated there is need address weaknesses threats large-scale globally representative populations leveraging approaches collaborative learning.

Language: Английский

Breath as a Biomarker: A Survey of Contact and Contactless Applications and Approaches in Respiratory Monitoring DOI

Almustapha Abdullahi Wakili,

Woosub Jung,

Babajide Asaju

et al.

Published: Jan. 1, 2025

Language: Английский

Citations

0
Early Diagnosis of Pneumonia and Chronic Obstructive Pulmonary Disease with a Smart Stethoscope with Cloud Server-Embedded Machine Learning in the Post-COVID-19 Era DOI Creative Commons

Direk Sueaseenak,

Peeravit Boonsat,

Suchada Tantisatirapong

et al.

Biomedicines, Journal Year: 2025, Volume and Issue: 13(2), P. 354 - 354

Published: Feb. 4, 2025

Background/Objectives: Respiratory diseases are common and result in high mortality, especially the elderly, with pneumonia chronic obstructive pulmonary disease (COPD). Auscultation of lung sounds using a stethoscope is crucial method for diagnosis, but it may require specialized training involvement pulmonologists. This study aims to assist medical professionals who non-pulmonologist doctors early screening COPD by developing smart cloud server-embedded machine learning diagnose sounds. Methods: The was developed Micro-Electro-Mechanical system (MEMS) microphone record mobile application then send them wirelessly server real-time classification. Results: model classifies into four categories: normal, pneumonia, COPD, other respiratory diseases. It achieved an accuracy 89%, sensitivity 89.75%, specificity 95%. In addition, testing healthy volunteers yielded 80% distinguishing normal diseased lungs. Moreover, performance comparison between two commercial auscultation stethoscopes showed comparable sound quality loudness results. Conclusions: holds great promise improving healthcare delivery post-COVID-19 era, offering probability most likely conditions diagnosis Its user-friendly design capabilities provide valuable resource delivering timely, evidence-based diagnoses, aiding treatment decisions, paving way more accessible care.

Language: Английский

Citations

0
Machine Learning-Based Methods for Detecting Respiratory Abnormalities Using Audio and Visual Analysis: A Review DOI Creative Commons

Sufyan Othman Zaben,

Wan Mohd Nazmee Wan Zainon, Ahmad H. Sabry

et al.

Results in Engineering, Journal Year: 2025, Volume and Issue: unknown, P. 104744 - 104744

Published: March 1, 2025

Language: Английский

Citations

0
Non-invasive diagnosis of lung diseases via multimodal feature extraction from breathing audio and chest dynamics DOI
Alyaa Hamel Sfayyih, Nasri Sulaiman, Ahmad H. Sabry

et al.

Computers in Biology and Medicine, Journal Year: 2025, Volume and Issue: 191, P. 110182 - 110182

Published: April 10, 2025

Language: Английский

Citations

0
Transforming pulmonary healthcare: the role of artificial intelligence in diagnosis and treatment DOI

Farzat Farha,

Sageer Abass,

Saba Khan

et al.

Expert Review of Respiratory Medicine, Journal Year: 2025, Volume and Issue: unknown

Published: April 10, 2025

Respiratory diseases like pneumonia, asthma, and COPD are major global health concerns, significantly impacting morbidity mortality rates worldwide. A selective search on PubMed, Google Scholar, ScienceDirect (up to 2024) focused AI in diagnosing treating respiratory conditions COPD. Studies were chosen for their relevance prediction models, AI-driven diagnostics, personalized treatments. This narrative review highlights technological advancements, clinical applications, challenges integrating into standard practice, with emphasis predictive tools, deep learning imaging, patient outcomes. Despite these significant remain fully pulmonary healthcare. The need large, diverse datasets train models is critical, concerns around data privacy, algorithmic transparency, potential biases must be carefully managed. Regulatory frameworks also evolve address the unique posed by However, continued research collaboration between technology developers, clinicians, policymakers, has revolutionize healthcare, ultimately leading more effective, efficient, care patients.

Language: Английский

Citations

0
ELW‐CNN: An extremely lightweight convolutional neural network for enhancing interoperability in colon and lung cancer identification using explainable AI DOI Creative Commons

Shaiful Ajam Opee,

Arifa Akter Eva,

Asaduzzaman Noor

et al.

Healthcare Technology Letters, Journal Year: 2025, Volume and Issue: 12(1)

Published: Jan. 1, 2025

Abstract Cancer is a condition in which cells the body grow uncontrollably, often forming tumours and potentially spreading to various areas of body. hazardous medical case history analysis. Every year, many people die cancer at an early stage. Therefore, it necessary accurately identify effectively treat save human lives. However, machine deep learning models are effective for identification. effectiveness these efforts limited by small dataset size, poor data quality, interclass changes between lung squamous cell carcinoma adenocarcinoma, difficulties with mobile device deployment, lack image individual‐level accuracy tests. To overcome difficulties, this study proposed extremely lightweight model using convolutional neural network that achieved 98.16% large colon individually 99.02% 99.40% cancer. The used only 70 thousand parameters, highly real‐time solutions. Explainability methods such as Grad‐CAM symmetric explanation highlight specific regions input affect decision model, helping potential challenges. will aid professionals developing automated accurate approach detecting types

Language: Английский

Citations

0
Breath as a Biomarker: A Survey of Contact and Contactless Applications and Approaches in Respiratory Monitoring DOI
A.L.M.U.S.T.A.P.H.A.A. WAKILI,

Babajide Asaju,

Woosub Jung

et al.

Smart Health, Journal Year: 2025, Volume and Issue: unknown, P. 100579 - 100579

Published: April 1, 2025

Language: Английский

Citations

0
Automated Interpretation of Lung Sounds by Deep Learning in Children With Asthma: Scoping Review and Strengths, Weaknesses, Opportunities, and Threats Analysis DOI Creative Commons
Isabelle Ruchonnet‐Metrailler, Johan N Siebert, Mary-Anne Hartley

et al.

Journal of Medical Internet Research, Journal Year: 2024, Volume and Issue: 26, P. e53662 - e53662

Published: Aug. 23, 2024

Background The interpretation of lung sounds plays a crucial role in the appropriate diagnosis and management pediatric asthma. Applying artificial intelligence (AI) to this task has potential better standardize assessment may even improve its predictive potential. Objective This study aims objectively review literature on AI-assisted auscultation for asthma provide balanced strengths, weaknesses, opportunities, threats. Methods A scoping sound analysis children with was conducted across 4 major scientific databases (PubMed, MEDLINE Ovid, Embase, Web Science), supplemented by gray search Google Scholar, identify relevant studies published from January 1, 2000, until May 23, 2023. strategy incorporated combination keywords related AI, pulmonary auscultation, children, quality eligible assessed using ChAMAI (Checklist Assessment Medical Artificial Intelligence). Results identified 7 out 82 (9%) be included through an academic search, while 11 250 (4.4%) were considered but not subsequent assessment. All had poor medium scores, mostly due absence external validation. Identified strengths improved accuracy AI allow prompt early diagnosis, personalized strategies, remote monitoring capabilities. Weaknesses heterogeneity between lack standardization data collection interpretation. Opportunities coordinated surveillance, growing sets, new ways collaboratively learning distributed data. Threats both generic field medical (loss interpretability) also specific use case, as clinicians might lose skill auscultation. Conclusions To achieve opportunities automated there is need address weaknesses threats large-scale globally representative populations leveraging approaches collaborative learning.

Language: Английский

Citations

2
Artificial intelligence and pneumonia: a rapidly evolving frontier DOI Creative Commons
Amy Sarah Ginsburg, Eric D. McCollum

The Lancet Global Health, Journal Year: 2023, Volume and Issue: 11(12), P. e1849 - e1850

Published: Nov. 10, 2023

The use of artificial intelligence is increasing and permeates many aspects our daily lives. Its application in medicine also increasing, supported by the increased digitalisation clinical data. COVID-19 pandemic accelerated innovation artificial-intelligence-assisted medical care, particularly around lung imaging respiratory sounds, which have applications for pneumonia.1Jia LL Zhao JX Pan NN et al.Artificial model on chest to diagnose other pneumonias: a systematic review meta-analysis.Eur J Radiol Open. 2022; 9100438Summary Full Text PDF PubMed Scopus (6) Google Scholar Machine-leaning technologies algorithms that enable identification patterns been used with radiography, CT, ultrasound, MRI as well cough sounds aid diagnosis pneumonia diseases decision-making. Using complex structure based multilayered convolutional neural networks large numbers parameters are trained massive amounts data, deep learning automates high-level feature extraction produce accurate insights predictions. diagnostic performance deep-learning models conditions has reported be equivalent or superior health-care professionals.2Liu X Faes L Kale AU al.A comparison against professionals detecting from imaging: meta-analysis.Lancet Digit Health. 2019; 1: e271-e297Summary (730) Pneumonia typically diagnosed basis non-specific symptoms physical examination findings, can lead missed incorrect diagnoses, treatments, complications, deaths. As valuable source prognostic information COVID-19, subject human operation interpretation resource training challenges. Artificial-intelligence-assisted automated diagnostics prognostics provide real-time analysis, interpretation, decision-making support potential address these For example, an externally validated algorithm triaging patients at fever clinics China suspected CT showed high accuracy diagnosing across populations varied prevalence.3Wang M Xia C Huang al.Deep learning-based triage analysis lesion burden COVID-19: retrospective study external validation.Lancet 2020; 2: e506-e515Summary (59) In another USA, radiography had better predicting progression critical illness than did data radiologist-derived severity scores.4Jiao Z Choi JW Halsey K al.Prognostication using x-rays data: study.Lancet 2021; 3: e286-e294Summary (0) If both extended pneumonia, illnesses, they help relieve constraints improve efficiency, workflow, outcomes where modalities available. Deep-learning being develop digital stethoscope assessment illnesses.5Sethi AK Muddaloor P Anvekar al.Digital pulmonology practice phonopulmography leveraging intelligence: future perspectives dual microwave acoustic sensing imaging.Sensors (Basel). 2023; 235514Crossref Scholar, 6Sfayyih AH Sabry Jameel SM al.Acoustic-based architectures disease diagnosis: comprehensive overview.Diagnostics 131748PubMed 7Sharan RV Rahimi-Ardabili H Detecting acute pediatric population sound features machine learning: review.Int Med Inform. 176105093Crossref Globally, most common patient-reported reason seeking health care. recognition coughs, breathing patterns.7Sharan 8Zhang Wang HS Zhou HY al.Real-world verification algorithm-assisted auscultation breath children.Front Pediatr. 9627337Google Often subjective, variable, open clinician, objective consistent evaluations facilitated could greatly enhance assessments diseases, including pneumonia. Given affects vulnerable populations, transform improving accessibility, speed, accuracy, reliability, generalisability, quality, effectiveness, delivery services outcomes. might more effective senses, smaller subtle visual differences lower higher frequency sounds. addition expanding access resource-constrained, rural, remote settings without highly providers specialised expertise experience, substantial cost savings.9Khanna Maindarkar MA Viswanathan V al.Economics healthcare: vs treatment.Healthcare 102493PubMed benefits include performance, democratising knowledge excellence, automating drudgery, allocating limited resources.10Price WN Risks remedies care.https://www.brookings.edu/articles/risks-and-remedies-for-artificial-intelligence-in-health-care/Date: Nov 14, 2019Date accessed: September 23, 2023Google However, several limitations challenges need overcome become powerful tool management These subjectiveness reference standards (ground truths), inadequate volume low-quality insufficient evaluation real-world settings, validate models, lack transparency adequate reporting regulatory challenges.2Liu 10Price Furthermore, existing emerging risks such system errors, biases, privacy security breaches cause patient harm increase inequities anticipated addressed. Although resource-constrained benefit most, harm. tools do not substitute strengthening systems. we commemorate yet World Day world stubbornly, remains deadly communicable rapidly evolving frontier, recognise persistent unmet improved diagnostics, prognostics, treatment options, encourage continued innovation, research, evidence-based approaches prevention while taking care ensure responsible, sustainable, inclusive development intelligence-assisted pneumonia-related technologies. ASG paid consultant Caption EDM holds grants Bill & Melinda Gates Foundation, US National Institutes Health, Agency International Development, Centers Disease Control Prevention, Thrasher Research Fund studies infection; receives funding Moderna syncytial virus prevention; scientific Sonavi Labs. This arrangement reviewed approved Johns Hopkins University accordance its conflict interest policies.

Language: Английский

Citations

5
An open auscultation dataset for machine learning-based respiratory diagnosis studies DOI Creative Commons
Guanyu Zhou, Chengjian Liu, Xiaoguang Li

et al.

JASA Express Letters, Journal Year: 2024, Volume and Issue: 4(5)

Published: May 1, 2024

Machine learning enabled auscultating diagnosis can provide promising solutions especially for prescreening purposes. The bottleneck its potential success is that high-quality datasets training are still scarce. An open auscultation dataset consists of samples and annotations from patients healthy individuals established in this work the respiratory studies with machine learning, which both scientific importance practical potential. A approach examined to showcase use new lung sound classifications different diseases. available public online.

Language: Английский

Citations

1