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

Transformers for rapid detection of airway stenosis and stridor

Upper airway stenosis is a potentially life-threatening condition involving the narrowing of the airway. In more severe cases, airway stenosis may be accompanied by stridor, a type of disordered breathing caused by turbulent airflow. Patients with airway stenosis have a higher risk of airway failure and additional precautions must be taken before medical interventions like intubation. However, stenosis and stridor are often misdiagnosed as other respiratory conditions like asthma/wheezing, worsening outcomes. This report presents a unified dataset containing recorded breathing tasks from patients with stridor and airway stenosis. Customized transformer-based models were also trained to perform stenosis and stridor detection tasks using low-cost data from multiple acoustic prompts recorded on common devices. These methods achieved AUC scores of 0.875 for stenosis detection and 0.864 for stridor detection, demonstrating the potential to add value as screening tools in real-world clinical workflows, particularly in high-volume settings like emergency departments.

Atypical pathogen community-acquired pneumonia: an analysis of clinical characteristics, drug treatment, and prognosis in the related patients

INTRODUCTION

Serious respiratory infections can occur in both in-hospital and out-of-hospital settings. These infections are known as community-acquired pneumonias (CAPs). Streptococcus pneumoniae and other microorganisms commonly cause atypical pneumonia. This study examined the clinical features, medication therapy, and prognosis of 85 cases of community-acquired pneumonia (CAP) caused by Mycoplasma pneumoniae (MPP) and Chlamydia psittaci(C. psittaci)neumoniae (CPP).

METHODS

A retrospective analysis was conducted at Shaoxing People's Hospital from July 2021 to August 2024, using targeted next-generation sequencing (tNGS) of bronchoalveolar lavage fluid (BALF). Patients were classified into the MPP group (54 patients) and the CPP group (31 patients). Compared with the control group, the CPP group had a significantly lower proportion of patients with a contact history of poultry and birds, a shorter length of hospital stay, and a lower percentage of severe pneumonia cases.

RESULTS

The MPP group demonstrated higher incidences of cough and sputum production; conversely, the occurrences of fever, fatigue, diminished appetite, and generalised myalgia were comparatively lower. The MPP group exhibited markedly diminished levels of neutrophils, C-reactive protein, procalcitonin, erythrocyte sedimentation rate, heparin-binding protein, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, direct bilirubin, pH, lactic acid, and D-dimer compared with the CPP group. In contrast, the MPP group had a markedly higher lymphocyte count, platelet count, albumin levels, as well as higher concentrations of blood sodium and blood chloride. The drug treatment regimens differed between the two groups, resulting in one unfavourable outcome within the MPP group.

CONCLUSION

In summary, fatigue, fever, and reduced appetite are more prominent symptoms in patients with CPP, whereas cough and sputum production are the primary manifestations of MPP. Pleural effusion is more prevalent in patients with CPP, Additionally, these patients also have increased inflammatory responses and decreased immune function.

Edge Computing System for Automatic Detection of Chronic Respiratory Diseases Using Audio Analysis

Chronic respiratory diseases affect people worldwide, but conventional diagnostic methods may not be accessible in remote locations far from population centers. Sounds from the human respiratory system have displayed potential in autonomously detecting these diseases using artificial intelligence (AI). This article outlines the development of an audio-based edge computing system that automatically detects chronic respiratory diseases (CRDs). The system utilizes machine learning (ML) techniques to analyze audio recordings of respiratory sounds (cough and breath) and classify the presence or absence of these diseases, using features such as Mel frequency cepstral coefficients (MFCC) and chromatic attributes (chromagram) to capture the relevant acoustic features of breath sounds. The system was trained and tested using a dataset of respiratory sounds collected from 86 individuals. Among them, 53 had chronic respiratory conditions, including asthma and chronic obstructive pulmonary disease (COPD), while the remaining 33 were healthy. The system's final evaluation was conducted with a group of 13 patients and 22 healthy individuals. Our approach demonstrated high sensitivity and specificity in the classification of sounds on edge devices, including smartphone and Raspberry Pi. Our best results for CRDs reached a sensitivity of 90.0%, a specificity of 93.55%, and a balanced accuracy of 91.75% for accurately identifying individuals with both healthy and diseased. These results showcase the potential of edge computing and machine learning systems in respiratory disease detection. We believe this system can contribute to developing efficient and cost-effective screening tools.

Automated Cough Analysis with Convolutional Recurrent Neural Network

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

Digital twins for chronic lung diseases

Digital twins have recently emerged in healthcare. They combine advances in cyber-physical systems, modelling and computation techniques, and enable a bidirectional flow of information between the physical and virtual entities. In respiratory medicine, progress in connected devices and artificial intelligence make it technically possible to obtain digital twins that allow real-time visualisation of a patient's respiratory health. Advances in respiratory system modelling also enable the development of digital twins that could be used to predict the effectiveness of different therapeutic approaches for a patient. For researchers, digital twins could lead to a better understanding of the gene-environment-time interactions involved in the development of chronic respiratory diseases. For clinicians and patients, they could facilitate personalised and timely medicine, by enabling therapeutic adaptations specific to each patient and early detection of disease progression. The objective of this review is to allow the reader to explore the concept of digital twins, their feasibility in respiratory medicine, their potential benefits and the challenges to their implementation.

An automatic cough counting method and system construction for portable devices

Introduction

Cough is a common symptom of respiratory diseases, and prolonged monitoring of cough can help assist doctors in making judgments about patients' conditions, among which cough frequency is an indicator that characterizes the state of the patient's lungs. Therefore, the aim of this paper is to design an automatic cough counting system to monitor the number of coughs per minute for a long period of time.

Methods

In this paper, a complete cough counting process is proposed, including denoising, segment extraction, eigenvalue calculation, recognition, and counting process; and a wearable automatic cough counting device containing acquisition and reception software. The design and construction of the algorithm is based on realistically captured cough-containing audio from 50 patients, combined with short-time features, and Meier cepstrum coefficients as features characterizing the cough.

Results

The accuracy, sensitivity, specificity, and F1 score of the method were 93.24%, 97.58%, 86.97%, and 94.47%, respectively, with a Kappa value of 0.9209, an average counting error of 0.46 counts for a 60-s speech segment, and an average runtime of 2.80 ± 2.27 s.

Discussion

This method improves the double threshold method in terms of the threshold and eigenvalues of the cough segments' sensitivity and has better performance in terms of accuracy, real-time performance, and computing speed, which can be applied to real-time cough counting and monitoring in small portable devices with limited computing power. The developed wearable portable automatic cough counting device and the accompanying host computer software application can realize the long-term monitoring of patients' coughing condition.

Artificial Intelligence-Driven Prognosis of Respiratory Mechanics: Forecasting Tissue Hysteresivity Using Long Short-Term Memory and Continuous Sensor Data

Tissue hysteresivity is an important marker for determining the onset and progression of respiratory diseases, calculated from forced oscillation lung function test data. This study aims to reduce the number and duration of required measurements by combining multivariate data from various sensing devices. We propose using the Forced Oscillation Technique (FOT) lung function test in both a low-frequency prototype and the commercial RESMON device, combined with continuous monitoring from the Equivital (EQV) LifeMonitor and processed by artificial intelligence (AI) algorithms. While AI and deep learning have been employed in various aspects of respiratory system analysis, such as predicting lung tissue displacement and respiratory failure, the prediction or forecasting of tissue hysteresivity remains largely unexplored in the literature. In this work, the Long Short-Term Memory (LSTM) model is used in two ways: (1) to estimate the hysteresivity coefficient η using heart rate (HR) data collected continuously by the EQV sensor, and (2) to forecast η values by first predicting the heart rate from electrocardiogram (ECG) data. Our methodology involves a rigorous two-hour measurement protocol, with synchronized data collection from the EQV, FOT, and RESMON devices. Our results demonstrate that LSTM networks can accurately estimate the tissue hysteresivity parameter η, achieving an R2 of 0.851 and a mean squared error (MSE) of 0.296 for estimation, and forecast η with an R2 of 0.883 and an MSE of 0.528, while significantly reducing the number of required measurements by a factor of three (i.e., from ten to three) for the patient. We conclude that our novel approach minimizes patient effort by reducing the measurement time and the overall ambulatory time and costs while highlighting the potential of artificial intelligence methods in respiratory monitoring.

Assessment of Pepper Robot's Speech Recognition System through the Lens of Machine Learning

Speech comprehension can be challenging due to multiple factors, causing inconvenience for both the speaker and the listener. In such situations, using a humanoid robot, Pepper, can be beneficial as it can display the corresponding text on its screen. However, prior to that, it is essential to carefully assess the accuracy of the audio recordings captured by Pepper. Therefore, in this study, an experiment is conducted with eight participants with the primary objective of examining Pepper's speech recognition system with the help of audio features such as Mel-Frequency Cepstral Coefficients, spectral centroid, spectral flatness, the Zero-Crossing Rate, pitch, and energy. Furthermore, the K-means algorithm was employed to create clusters based on these features with the aim of selecting the most suitable cluster with the help of the speech-to-text conversion tool Whisper. The selection of the best cluster is accomplished by finding the maximum accuracy data points lying in a cluster. A criterion of discarding data points with values of WER above 0.3 is imposed to achieve this. The findings of this study suggest that a distance of up to one meter from the humanoid robot Pepper is suitable for capturing the best speech recordings. In contrast, age and gender do not influence the accuracy of recorded speech. The proposed system will provide a significant strength in settings where subtitles are required to improve the comprehension of spoken statements.

Syndromic surveillance of population-level COVID-19 burden with cough monitoring in a hospital emergency waiting room

Syndromic surveillance is an effective tool for enabling the timely detection of infectious disease outbreaks and facilitating the implementation of effective mitigation strategies by public health authorities. While various information sources are currently utilized to collect syndromic signal data for analysis, the aggregated measurement of cough, an important symptom for many illnesses, is not widely employed as a syndromic signal. With recent advancements in ubiquitous sensing technologies, it becomes feasible to continuously measure population-level cough incidence in a contactless, unobtrusive, and automated manner. In this work, we demonstrate the utility of monitoring aggregated cough count as a syndromic indicator to estimate COVID-19 cases. In our study, we deployed a sensor-based platform (Syndromic Logger) in the emergency room of a large hospital. The platform captured syndromic signals from audio, thermal imaging, and radar, while the ground truth data were collected from the hospital's electronic health record. Our analysis revealed a significant correlation between the aggregated cough count and positive COVID-19 cases in the hospital (Pearson correlation of 0.40, p-value < 0.001). Notably, this correlation was higher than that observed with the number of individuals presenting with fever (ρ = 0.22, p = 0.04), a widely used syndromic signal and screening tool for such diseases. Furthermore, we demonstrate how the data obtained from our Syndromic Logger platform could be leveraged to estimate various COVID-19-related statistics using multiple modeling approaches. Aggregated cough counts and other data, such as people density collected from our platform, can be utilized to predict COVID-19 patient visits related metrics in a hospital waiting room, and SHAP and Gini feature importance-based metrics showed cough count as the important feature for these prediction models. Furthermore, we have shown that predictions based on cough counting outperform models based on fever detection (e.g., temperatures over 39°C), which require more intrusive engagement with the population. Our findings highlight that incorporating cough-counting based signals into syndromic surveillance systems can significantly enhance overall resilience against future public health challenges, such as emerging disease outbreaks or pandemics.

Respiratory Diseases Diagnosis Using Audio Analysis and Artificial Intelligence: A Systematic Review

Respiratory diseases represent a significant global burden, necessitating efficient diagnostic methods for timely intervention. Digital biomarkers based on audio, acoustics, and sound from the upper and lower respiratory system, as well as the voice, have emerged as valuable indicators of respiratory functionality. Recent advancements in machine learning (ML) algorithms offer promising avenues for the identification and diagnosis of respiratory diseases through the analysis and processing of such audio-based biomarkers. An ever-increasing number of studies employ ML techniques to extract meaningful information from audio biomarkers. Beyond disease identification, these studies explore diverse aspects such as the recognition of cough sounds amidst environmental noise, the analysis of respiratory sounds to detect respiratory symptoms like wheezes and crackles, as well as the analysis of the voice/speech for the evaluation of human voice abnormalities. To provide a more in-depth analysis, this review examines 75 relevant audio analysis studies across three distinct areas of concern based on respiratory diseases' symptoms: (a) cough detection, (b) lower respiratory symptoms identification, and (c) diagnostics from the voice and speech. Furthermore, publicly available datasets commonly utilized in this domain are presented. It is observed that research trends are influenced by the pandemic, with a surge in studies on COVID-19 diagnosis, mobile data acquisition, and remote diagnosis systems.

An AI-Enabled Bias-Free Respiratory Disease Diagnosis Model Using Cough Audio

Cough-based diagnosis for respiratory diseases (RDs) using artificial intelligence (AI) has attracted considerable attention, yet many existing studies overlook confounding variables in their predictive models. These variables can distort the relationship between cough recordings (input data) and RD status (output variable), leading to biased associations and unrealistic model performance. To address this gap, we propose the Bias-Free Network (RBF-Net), an end-to-end solution that effectively mitigates the impact of confounders in the training data distribution. RBF-Net ensures accurate and unbiased RD diagnosis features, emphasizing its relevance by incorporating a COVID-19 dataset in this study. This approach aims to enhance the reliability of AI-based RD diagnosis models by navigating the challenges posed by confounding variables. A hybrid of a Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks is proposed for the feature encoder module of RBF-Net. An additional bias predictor is incorporated in the classification scheme to formulate a conditional Generative Adversarial Network (c-GAN) that helps in decorrelating the impact of confounding variables from RD prediction. The merit of RBF-Net is demonstrated by comparing classification performance with a State-of-The-Art (SoTA) Deep Learning (DL) model (CNN-LSTM) after training on different unbalanced COVID-19 data sets, created by using a large-scale proprietary cough data set. RBF-Net proved its robustness against extremely biased training scenarios by achieving test set accuracies of 84.1%, 84.6%, and 80.5% for the following confounding variables-gender, age, and smoking status, respectively. RBF-Net outperforms the CNN-LSTM model test set accuracies by 5.5%, 7.7%, and 8.2%, respectively.

Identifying Existing Evidence to Potentially Develop a Machine Learning Diagnostic Algorithm for Cough in Primary Care Settings: Scoping Review

BACKGROUND

Primary care is known to be one of the most complex health care settings because of the high number of theoretically possible diagnoses. Therefore, the process of clinical decision-making in primary care includes complex analytical and nonanalytical factors such as gut feelings and dealing with uncertainties. Artificial intelligence is also mandated to offer support in finding valid diagnoses. Nevertheless, to translate some aspects of what occurs during a consultation into a machine-based diagnostic algorithm, the probabilities for the underlying diagnoses (odds ratios) need to be determined.

OBJECTIVE

Cough is one of the most common reasons for a consultation in general practice, the core discipline in primary care. The aim of this scoping review was to identify the available data on cough as a predictor of various diagnoses encountered in general practice. In the context of an ongoing project, we reflect on this database as a possible basis for a machine-based diagnostic algorithm. Furthermore, we discuss the applicability of such an algorithm against the background of the specifics of general practice.

METHODS

The PubMed, Scopus, Web of Science, and Cochrane Library databases were searched with defined search terms, supplemented by the search for gray literature via the German Journal of Family Medicine until April 20, 2023. The inclusion criterion was the explicit analysis of cough as a predictor of any conceivable disease. Exclusion criteria were articles that did not provide original study results, articles in languages other than English or German, and articles that did not mention cough as a diagnostic predictor.

RESULTS

In total, 1458 records were identified for screening, of which 35 articles met our inclusion criteria. Most of the results (11/35, 31%) were found for chronic obstructive pulmonary disease. The others were distributed among the diagnoses of asthma or unspecified obstructive airway disease, various infectious diseases, bronchogenic carcinoma, dyspepsia or gastroesophageal reflux disease, and adverse effects of angiotensin-converting enzyme inhibitors. Positive odds ratios were found for cough as a predictor of chronic obstructive pulmonary disease, influenza, COVID-19 infections, and bronchial carcinoma, whereas the results for cough as a predictor of asthma and other nonspecified obstructive airway diseases were inconsistent.

CONCLUSIONS

Reliable data on cough as a predictor of various diagnoses encountered in general practice are scarce. The example of cough does not provide a sufficient database to contribute odds to a machine learning-based diagnostic algorithm in a meaningful way.

Screening COVID-19 by Swaasa AI platform using cough sounds: a cross-sectional study

The Advent of Artificial Intelligence (AI) has led to the use of auditory data for detecting various diseases, including COVID-19. SARS-CoV-2 infection has claimed more than six million lives to date and therefore, needs a robust screening technique to control the disease spread. In the present study we created and validated the Swaasa AI platform, which uses the signature cough sound and symptoms presented by patients to screen and prioritize COVID-19 patients. We collected cough data from 234 COVID-19 suspects to validate our Convolutional Neural Network (CNN) architecture and Feedforward Artificial Neural Network (FFANN) (tabular features) based algorithm. The final output from both models was combined to predict the likelihood of having the disease. During the clinical validation phase, our model showed a 75.54% accuracy rate in detecting the likely presence of COVID-19, with 95.45% sensitivity and 73.46% specificity. We conducted pilot testing on 183 presumptive COVID subjects, of which 58 were truly COVID-19 positive, resulting in a Positive Predictive Value of 70.73%. Due to the high cost and technical expertise required for currently available rapid screening methods, there is a need for a cost-effective and remote monitoring tool that can serve as a preliminary screening method for potential COVID-19 subjects. Therefore, Swaasa would be highly beneficial in detecting the disease and could have a significant impact in reducing its spread.

HEAR4Health: a blueprint for making computer audition a staple of modern healthcare

Recent years have seen a rapid increase in digital medicine research in an attempt to transform traditional healthcare systems to their modern, intelligent, and versatile equivalents that are adequately equipped to tackle contemporary challenges. This has led to a wave of applications that utilise AI technologies; first and foremost in the fields of medical imaging, but also in the use of wearables and other intelligent sensors. In comparison, computer audition can be seen to be lagging behind, at least in terms of commercial interest. Yet, audition has long been a staple assistant for medical practitioners, with the stethoscope being the quintessential sign of doctors around the world. Transforming this traditional technology with the use of AI entails a set of unique challenges. We categorise the advances needed in four key pillars: Hear, corresponding to the cornerstone technologies needed to analyse auditory signals in real-life conditions; Earlier, for the advances needed in computational and data efficiency; Attentively, for accounting to individual differences and handling the longitudinal nature of medical data; and, finally, Responsibly, for ensuring compliance to the ethical standards accorded to the field of medicine. Thus, we provide an overview and perspective of HEAR4Health: the sketch of a modern, ubiquitous sensing system that can bring computer audition on par with other AI technologies in the strive for improved healthcare systems.

Automatic Classification of Hospital Settings through Artificial Intelligence

Modern hospitals have to meet requirements from national and international institutions in order to ensure hygiene, quality and organisational standards. Moreover, a hospital must be flexible and adaptable to new delivery models for healthcare services. Various hospital monitoring tools have been developed over the years, which allow for a detailed picture of the effectiveness and efficiency of the hospital itself. Many of these systems are based on database management systems (DBMSs), building information modelling (BIM) or geographic information systems (GISs). This work presents an automatic recognition system for hospital settings that integrates these tools. Three alternative proposals were analysed in terms of the construction of the system: the first was based on the use of general models that are present on the cloud for the classification of images; the second consisted of the creation of a customised model and referred to the Clarifai Custom Model service; the third used an object recognition software that was developed by Facebook AI Research combined with a random forest classifier. The obtained results were promising. The customised model almost always classified the photos according to the correct intended use, resulting in a high percentage of confidence of up to 96%. Classification using the third tool was excellent when considering a limited number of hospital settings, with a peak accuracy of higher than 99% and an area under the ROC curve (AUC) of one for specific classes. As expected, increasing the number of room typologies to be discerned negatively affected performance.