Lung Sound Classification Using Co-Tuning and Stochastic Normalization

IEEE Transactions on Biomedical Engineering, Год журнала: 2022, Номер 69(9), С. 2872 - 2882

Опубликована: Март 7, 2022

Computational methods for lung sound analysis are beneficial computer-aided diagnosis support, storage and monitoring in critical care. In this paper, we use pre-trained ResNet models as backbone architectures classification of adventitious sounds respiratory diseases. The learned representation the model is transferred by using vanilla fine-tuning, co-tuning, stochastic normalization combination co-tuning techniques. Furthermore, data augmentation both time domain time-frequency used to account class imbalance ICBHI our multi-channel dataset. Additionally, introduce spectrum correction variations recording device properties on Empirically, proposed systems mostly outperform all state-of-the-art diseases datasets.

Язык: Английский

---

An automated diagnosis and classification of COVID-19 from chest CT images using a transfer learning-based convolutional neural network

Computers in Biology and Medicine, Год журнала: 2022, Номер 144, С. 105383 - 105383

Опубликована: Март 10, 2022

Язык: Английский

---

Reshaping the healthcare world by AI-integrated wearable sensors following COVID-19

Chemical Engineering Journal, Год журнала: 2025, Номер 505, С. 159478 - 159478

Опубликована: Янв. 11, 2025

Язык: Английский

---

Towards using cough for respiratory disease diagnosis by leveraging Artificial Intelligence: A survey

Informatics in Medicine Unlocked, Год журнала: 2022, Номер 29, С. 100832 - 100832

Опубликована: Янв. 1, 2022

Cough acoustics contain multitudes of vital information about pathomorphological alterations in the respiratory system. Reliable and accurate detection cough events by investigating underlying latent features disease diagnosis can play an indispensable role revitalizing healthcare practices. The recent application Artificial Intelligence (AI) advances ubiquitous computing for prediction has created auspicious trend myriad future possibilities medical domain. In particular, there is expeditiously emerging Machine learning (ML) Deep Learning (DL)-based diagnostic algorithms exploiting signatures. enormous body literature on cough-based AI demonstrate that these models a significant detecting onset specific disease. However, it pertinent to collect from all relevant studies exhaustive manner experts scientists analyze decisive AI/ML. This survey offers comprehensive overview data-driven ML/DL preliminary frameworks, along with detailed list features. We investigate mechanism causes modalities. also customized monitoring application, their AI- powered recognition algorithms. Challenges prospective research directions develop practical, robust, solutions are discussed detail.

Язык: Английский

---

A novel explainable COVID-19 diagnosis method by integration of feature selection with random forest

Informatics in Medicine Unlocked, Год журнала: 2022, Номер 30, С. 100941 - 100941

Опубликована: Янв. 1, 2022

Several Artificial Intelligence-based models have been developed for COVID-19 disease diagnosis. In spite of the promise artificial intelligence, there are very few which bridge gap between traditional human-centered diagnosis and potential future machine-centered Under concept human-computer interaction design, this study proposes a new explainable intelligence method that exploits graph analysis feature visualization optimization purpose from blood test samples. model, an decision forest classifier is employed to classification based on routinely available patient data. The approach enables clinician use tree guide explainability interpretability prediction model. By utilizing novel selection phase, proposed model will not only improve accuracy but decrease execution time as well.

Язык: Английский

---

Detection of COVID-19 in smartphone-based breathing recordings: A pre-screening deep learning tool

PLoS ONE, Год журнала: 2022, Номер 17(1), С. e0262448 - e0262448

Опубликована: Янв. 13, 2022

This study was sought to investigate the feasibility of using smartphone-based breathing sounds within a deep learning framework discriminate between COVID-19, including asymptomatic, and healthy subjects. A total 480 (240 shallow 240 deep) were obtained from publicly available database named Coswara. These recorded by 120 COVID-19 subjects via smartphone microphone through website application. proposed herein that relies on hand-crafted features extracted original recordings mel-frequency cepstral coefficients (MFCC) as well deep-activated learned combination convolutional neural network bi-directional long short-term memory units (CNN-BiLSTM). The statistical analysis patient profiles has shown significant difference (p-value: 0.041) for ischemic heart disease Analysis normal distribution combined MFCC values showed tended have is skewed more towards right side zero mean (shallow: 0.59±1.74, deep: 0.65±4.35, p-value: <0.001). In addition, approach had an overall discrimination accuracy 94.58% 92.08% recordings, respectively. Furthermore, it detected successfully with maximum sensitivity 94.21%, specificity 94.96%, area under receiver operating characteristic (AUROC) curves 0.90. Among participants, asymptomatic (18 subjects) 100.00% 88.89% recordings. paves way utilizing purpose detection. observations found in this promising suggest effective pre-screening tool alongside current reverse-transcription polymerase chain reaction (RT-PCR) assay. It can be considered early, rapid, easily distributed, time-efficient, almost no-cost diagnosis technique complying social distancing restrictions during pandemic.

Язык: Английский

---

Project Achoo: A Practical Model and Application for COVID-19 Detection From Recordings of Breath, Voice, and Cough

IEEE Journal of Selected Topics in Signal Processing, Год журнала: 2022, Номер 16(2), С. 175 - 187

Опубликована: Янв. 13, 2022

The COVID-19 pandemic created significant interest and demand for infection detection monitoring solutions. In this paper, we propose a machine learning method to quickly detect using audio recordings made on consumer devices. approach combines signal processing noise removal methods with an ensemble of fine-tuned deep networks enables COVID coughs. We have also developed deployed mobile application that uses symptoms checker together voice, breath, cough signals infection. showed robust performance both openly sourced datasets the noisy data collected during beta testing by end users.

Язык: Английский

---

COVID-19 disease diagnosis with light-weight CNN using modified MFCC and enhanced GFCC from human respiratory sounds

The European Physical Journal Special Topics, Год журнала: 2022, Номер 231(18-20), С. 3329 - 3346

Опубликована: Янв. 24, 2022

Язык: Английский

---

Wide‐Bandwidth Nanocomposite‐Sensor Integrated Smart Mask for Tracking Multiphase Respiratory Activities

Advanced Science, Год журнала: 2022, Номер 9(31)

Опубликована: Авг. 23, 2022

Abstract Wearing masks has been a recommended protective measure due to the risks of coronavirus disease 2019 (COVID‐19) even in its coming endemic phase. Therefore, deploying “smart mask” monitor human physiological signals is highly beneficial for personal and public health. This work presents smart mask integrating an ultrathin nanocomposite sponge structure‐based soundwave sensor (≈400 µm), which allows high sensitivity wide‐bandwidth dynamic pressure range, i.e., capable detecting various respiratory sounds breathing, speaking, coughing. Thirty‐one subjects test recording their activities. Machine/deep learning methods, support vector machine convolutional neural networks, are used recognize these activities, show average macro‐recalls ≈95% both individual generalized models. With rich high‐frequency (≈4000 Hz) information recorded, two‐/tri‐phase coughs can be mapped while speaking words identified, demonstrating that applicable as daily wearable Internet Things (IoT) device identification, voice interaction tool, etc. future. bridges technological gap between ultra‐lightweight but response material fabrication, signal transduction processing, machining/deep demonstrate potential applications continual health monitoring life.

Язык: Английский

---

Diagnosing COVID-19 using artificial intelligence: a comprehensive review

Network Modeling Analysis in Health Informatics and Bioinformatics, Год журнала: 2022, Номер 11(1)

Опубликована: Июль 12, 2022

Abstract In early March 2020, the World Health Organization (WHO) proclaimed novel COVID-19 as a global pandemic. The coronavirus went on to be life-threatening infection and is still wreaking havoc all around globe. Though vaccines have been rolled out, section of population (the elderly people with comorbidities) succumb this deadly illness. Hence, it imperative diagnose prevent potential severe prognosis. This contagious disease usually diagnosed using conventional technique called Reverse Transcription Polymerase Chain Reaction (RT-PCR). However, procedure leads number wrong false-negative results. Moreover, might also not newer variants mutating virus. Artificial Intelligence has one most widely discussed topics in recent years. It used tackle various issues across multiple domains modern world. extensive review, applications detection modalities such CT-Scans, X-rays, Cough sounds, MRIs, ultrasound clinical markers are explored depth. review provides data enthusiasts broader health community complete assessment current state-of-the-art approaches diagnosing COVID-19. key future directions provided for upcoming researchers.

Язык: Английский

---