Detection of tuberculosis using cough audio analysis: a deep learning approach with capsule networks DOI Creative Commons
Sakthi Jaya Sundar Rajasekar,

Anu Rithiga Balaraman,

Deepa Varnika Balaraman

и другие.

Discover Artificial Intelligence, Год журнала: 2024, Номер 4(1)

Опубликована: Ноя. 4, 2024

Tuberculosis (TB) is a widespread infectious disease that requires early detection for effective treatment and control. This study aims to improve TB using cough audio analysis, comparing the performance of capsule networks other deep learning models. We used recordings from 1105 individuals with new or worsening at least two weeks, totaling 9772 recordings. These were processed into spectral images, HOG features extracted. Various models, including Capsule Networks + FCNN, CNN, VGG16, ResNet50 trained evaluated. FCNN achieved best an accuracy 0.97, sensitivity 0.98, specificity 0.96, F1 score precision outperforming attribute due model's ability learn complex images. concludes are more than typical CNN-based models in diagnosing audio. suggests advanced frameworks could significantly enhance screening accuracy, especially resource-limited areas.

Язык: Английский

COVID-19 Detection from Optimized Features of Breathing Audio Signals Using Explainable Ensemble Machine Learning DOI Creative Commons
Shahnaz Sultana, A. B. M. Aowlad Hossain, Jahangir Alam

и другие.

Results in Control and Optimization, Год журнала: 2025, Номер unknown, С. 100538 - 100538

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

0
Early Detection and Monitoring of Respiratory Disorders using LASSO Regression on PPG Signals with Elephant Search Optimization DOI Open Access
M. S., Harikumar Rajaguru,

M. Kalaiyarasi

и другие.

Journal of Innovative Image Processing, Год журнала: 2025, Номер 7(1), С. 74 - 96

Опубликована: Март 1, 2025

Early diagnosis is the need of hour in treatment respiratory-related health conditions. This study presents a novel method for monitoring respiratory disorders by applying Least Absolute Shrinkage and Selection Operator (LASSO) regression model to Photoplethysmography (PPG) signals. By analyzing variations PPG waveform, partial pressure carbon dioxide (PCO₂) signal extracted monitor breathing patterns. The PCO₂ provides critical insights into dynamics, enabling identification irregular rates airflow obstructions. Using LASSO regression, most relevant features from signals are selected, reducing dimensionality improving prediction accuracy. proposed approach offers cost-effective non-invasive solution evaluating health, making it suitable both clinical non-clinical settings. A comprehensive performance analysis demonstrates efficacy regression-based diagnosing To evaluate its performance, five machine learning classifiers were employed: Linear Regression, Bayesian Discriminant Analysis (BLDA), k-Nearest Neighbors (k-NN) with weighted voting, Expectation-Maximization (EM) Logistic Elephant Search Optimization (ESO). results highlight potential this improve healthcare early detection management disorders. Optimization, combined reduction, achieves 95.12% accuracy value, 95% F1 score, 0.90% MCC 4.87% error rate, 90.47% Jaccard metrics, 90% CSI.

Язык: Английский

Процитировано

0
COPDVD: Automated classification of chronic obstructive pulmonary disease on a new collected and evaluated voice dataset DOI Creative Commons
Alper Idrisoglu, Ana Luiza Dallora, Abbas Cheddad

и другие.

Artificial Intelligence in Medicine, Год журнала: 2024, Номер 156, С. 102953 - 102953

Опубликована: Авг. 15, 2024

Chronic obstructive pulmonary disease (COPD) is a severe condition affecting millions worldwide, leading to numerous annual deaths. The absence of significant symptoms in its early stages promotes high underdiagnosis rates for the affected people. Besides function failure, another harmful problem COPD systemic effects, e.g., heart failure or voice distortion. However, effects might provide valuable information detection. In other words, caused by could be helpful detect stages.

Язык: Английский

Процитировано

2
GAN-Enhanced Vocal Biomarker Analysis for Respiratory Health Assessment DOI Open Access

Prof. Shweta Bhelonde,

Abhinav Pandey,

M. Rahul Surya

и другие.

International Journal of Advanced Research in Science Communication and Technology, Год журнала: 2024, Номер unknown, С. 583 - 595

Опубликована: Июнь 14, 2024

Nearly two centuries ago, people became aware that various diseases, such as the common cold, asthma, Alzheimer's, and psychological disorders, manifest changes in a human voice. The recent emergence of virus known "COVID-19" has claimed millions lives due to delayed detection infected individuals. Traditional medical techniques for are time-consuming costly. However, advancements Artificial Intelligence (AI) offer remote diagnosis analysing identifying diseases cause variations evolution machine learning provides numerous extract meaningful information from vocal biomarkers. This study explores innovative enhance analysis biomarkers, emphasizing Generative Adversarial Networks (GANs) assessing respiratory diseases. end goal is improve performance by utilizing synthetic data training purposes. Subsequently, models employed analyze real-time detecting illnesses. Comparing different algorithms gives us better understanding their capabilities drawbacks

Язык: Английский

Процитировано

0
Recognition of Patient Gender: A Machine Learning Preliminary Analysis Using Heart Sounds from Children and Adolescents DOI

Rodrigo M Carrillo-Larco

Pediatric Cardiology, Год журнала: 2024, Номер unknown

Опубликована: Июнь 27, 2024

Язык: Английский

Процитировано

0
Automated Cough Analysis with Convolutional Recurrent Neural Network DOI Creative Commons
Yiping Wang,

Mustafaa Wahab,

Tianqi Hong

и другие.

Bioengineering, Год журнала: 2024, Номер 11(11), С. 1105 - 1105

Опубликована: Ноя. 1, 2024

Chronic cough is associated with several respiratory diseases and a significant burden on physical, social, psychological health. Non-invasive, real-time, continuous, quantitative monitoring tools are highly desired to assess severity, the effectiveness of treatment, monitor disease progression in clinical practice research. There currently limited quantitatively measure spontaneous coughs daily living settings trials practice. In this study, we developed machine learning model for detection classification sounds. Mel spectrograms utilized as key feature representation capture temporal spectral characteristics coughs. We applied approach automate analysis using 300 h audio recordings from challenge studies conducted lab setting. A number algorithms were studied compared, including decision tree, support vector machine, k-nearest neighbors, logistic regression, random forest, neural network. identified that dataset, CRNN most effective method, reaching 98% accuracy identifying individual data. These findings provide insights into strengths limitations various algorithms, highlighting potential CRNNs analyzing complex patterns. This research demonstrates network models fully automated monitoring. The requires validation detecting patients refractory chronic real-life

Язык: Английский

Процитировано

0
Detection of tuberculosis using cough audio analysis: a deep learning approach with capsule networks DOI Creative Commons
Sakthi Jaya Sundar Rajasekar,

Anu Rithiga Balaraman,

Deepa Varnika Balaraman

и другие.

Discover Artificial Intelligence, Год журнала: 2024, Номер 4(1)

Опубликована: Ноя. 4, 2024

Tuberculosis (TB) is a widespread infectious disease that requires early detection for effective treatment and control. This study aims to improve TB using cough audio analysis, comparing the performance of capsule networks other deep learning models. We used recordings from 1105 individuals with new or worsening at least two weeks, totaling 9772 recordings. These were processed into spectral images, HOG features extracted. Various models, including Capsule Networks + FCNN, CNN, VGG16, ResNet50 trained evaluated. FCNN achieved best an accuracy 0.97, sensitivity 0.98, specificity 0.96, F1 score precision outperforming attribute due model's ability learn complex images. concludes are more than typical CNN-based models in diagnosing audio. suggests advanced frameworks could significantly enhance screening accuracy, especially resource-limited areas.

Язык: Английский

Процитировано

0