Abstract
Background. Cough is one of the most common clinical manifestations of COVID-19. Objective of our study was to perform analysis of cough duration, energy and sound frequency in COVID-19 patients, compared to induced cough in controls and the cough in asthma and chronic obstructive pulmonary disease (COPD) patients. Methods. The following characteristics of cough sounds were obtained: duration, Q coefficient - low/medium-frequency energy (60–600 Hz) to high-frequency energy (600–6000 Hz) ratio, and frequency of maximum sound energy. The cough was divided into three phases and assessment of characteristics was applied to the entire coughing act and to each phase separately in controls and patients with COVID-19, asthma and COPD. Results. The cough sounds of COVID-19 patients were characterized by a shorter duration, a predominance of high-frequency energy and higher maximum frequency of the energy, compared with the induced cough of controls. However, the frequencies of the maximum sound energy of the individual cough phases did not differ significantly, as did the duration of the first phase. In addition, the significant differences were demonstrated in some time-frequency parameters of cough sounds in the patients with asthma and COPD as compared to COVID-19 patients. Conclusion. Therefore, we have shown the distinction between the cough characteristics of COVID-19 patients compared to controls and patients with asthma or COPD.
Keywords: COVID-19, Cough sound duration, Cough sound energy, Cough sound frequency
Introduction
Coronavirus disease 2019 (COVID-19) caused the death of millions of people and despite the active immunization of the population it continues to spread. On January 30, 2020, WHO declared a global health emergency as the disease was actively spreading in many countries [1, 2].
COVID-19 is an infectious disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) from the Coronaviridae family that affects the upper and lower respiratory tract and lung parenchyma [3, 4]. The clinical picture of the disease varies from asymptomatic forms to severe pneumonia with fatal outcomes. The most common symptoms of the disease are fever, cough, shortness of breath, fatigue, and chest tightness [5]. In addition to the respiratory symptoms, some patients describe problems of the cardiovascular system, gastrointestinal tract, liver, kidneys, central nervous system, etc [6].
A non-productive cough is one of the main symptoms for seeking medical help. It also causes severe discomfort to the patient and people around, and contributes to the spread of infection [7].
There is a range of methods for the cough assessment. One of the objective methods among them is the analysis of cough sound. Analysis of cough sound is considered to be a promising non-invasive diagnostic method. It allows evaluating the cough duration and the distribution of sound energy by frequency. The characteristics of cough differ in various diseases of the respiratory system [8–11].
Thus, the available literature contains data on the characteristics of the cough sounds in patients with various respiratory diseases. However, at the moment there is little information about cough sounds in patients with COVID-19.
Therefore, the purpose of our study was to conduct an analysis of cough duration, energy and sound frequency in COVID-19 patients in comparison with induced cough in control individuals without COVID-19 and the cough of patients with other respiratory diseases, such as asthma and chronic obstructive pulmonary disease (COPD).
Methods
Study design
This was a prospective cohort study conducted in a blinded-fashion manner in the COVID-19 units of Voronezh State Medical University. We compared cough sounds characteristics in patients with COVID-19 and non-COVID-19 volunteers, asthma and COPD patients. The study was conducted in accordance with the principles of the Declaration of Helsinki of the World Medical Association. The study was approved by the ethics committee of the institution. All the subjects gave the informed consent. We would like to mention that some data were previously published (controls and COVID-19 patients) in the local journal Pulmonologiya in Russian language and we have obtained the full permission to use the data [12].
Patients
All the patients included in our study (main group) were admitted with SARS‑CoV‑2 infection within 5 days of first symptoms onset in COVID-19 units of a university-affiliated hospital and were diagnosed with SARS‑CoV‑2 infection by detecting viral RNA in respiratory samples by reverse transcription‑polymerase chain reaction. We enrolled patients aged ≥ 18 years with SARS-CoV-2 infection with radiological findings compatible with COVID-19 pneumonia, without acute hypoxemic respiratory failure (oxygen saturation (SpO2) ≥ 94%). The exclusion criteria were the need for endotracheal intubation, unstable hemodynamics (need for catecholamine and/or life-threatening rhythm abnormalities), ≥ 25% of lung involvement according to chest computed tomography (CT), chronic respiratory diseases, and the body mass index (BMI) ≥ 30 kg/m2 (since higher values may affect the results of cough sounds analysis).
Non-COVID-19 volunteers (comparison group) were aged ≥ 18 years, without respiratory symptoms, smoking history or chronic respiratory disease, without allergy to citric acid and with BMI < 30 kg/m2. In addition, in order to compare and identify possible differences in the cough sounds characteristics, we examined patients with uncontrolled or partly controlled asthma, as well as COPD group D during the exacerbation.
Analysis of cough sounds
Cough sounds were recorded from pre-instructed patients using a microphone connected to the sound card input. The microphone had the following characteristics: bandwidth − 60–24,000 Hz, impedance − 300 ohms, sensitivity − 90 dB. The subjects were trained to correctly perform the coughing maneuver so that the amplitude of the sound wave did not differ significantly. The patient was seated at a table with a microphone on it. Using a special bracket, it was placed at a distance of 15–20 cm from the patient’s face parallel to the floor, which made it possible to record under the same conditions. Filters with a bandwidth of 60 to 6000 Hz (48 dB/octave; Butterworth) were also used to minimize the effects of noise and prevent aliasing. Spontaneous coughing has been recorded in COVID-19 patients. In non-COVID-19 control individuals, cough was induced by inhalation of a citric acid solution at a concentration of 20 g/L, after which it was recorded [13].
Analysis of cough sounds recorded in a free acoustic field was performed using the Sound Forge 15 computer program (MAGIX Software GmbH., Germany). Cough cascades (if any) were divided into separate coughing acts. The volume was normalized to 6 dB. The sampling frequency was 48,000 Hz. Each cough was divided into three phases as shown in the Fig. 1.
Fig. 1.
Visual separation of the cough audiogram into phases
Analysis of cough sounds was performed using the fast Fourier transform algorithm. The following time-frequency parameters were estimated:
Duration of cough (T) and each phase separately (T1, T2, T3), ms.
The ratio of the low and medium frequencies (60–600 Hz) energy to the high frequencies (600–6000 Hz) energy of the entire cough (Q) and each phase separately (Q1, Q2, Q3).
The frequency of the maximum sound energy of the entire cough (Fmax) and each phase separately (Fmax1, Fmax2, Fmax3), Hz
Statistical analysis
Mathematical and statistical processing of the obtained data was conducted using the STATGRAPHICS Centurion 18 software package (Statgraphics Technologies, Inc., USA). Using the normalized kurtosis and skewness coefficients, the normality of the distribution was estimated. Since the distribution of the variant was not normal, numerical measures of cough parameters were presented as a median, 25% and 75% percentiles (lower and upper quartiles) were indicated in brackets. To compare the two samples in terms of quantitative indicators, the Mann-Whitney U-test was used. At p < 0.05, the differences were considered statistically significant.
Results
218 consecutive patients with COVID-19 and 60 non-COVID-19 volunteers (controls), 52 asthma and 60 COPD patients were enrolled in our study. The baseline characteristics of patients are presented in the Table 1. The time between the symptom onset and analysis of cough sounds was 7.5 (5–10) days. The type of cough of our patients was predominantly dry.
Table 1.
Baseline characteristics of study groups
| Characteristics | COVID-19 patients (n = 218) | Control (n = 60) | Asthma (n = 52) | COPD (n = 60) |
|---|---|---|---|---|
| Demographics | ||||
| Male, n (%) | 106 (48.6) | 30 (50) | 23 (44.2) | 30 (50) |
| Female, n (%) | 112 (51.4) | 30 (50) | 29 (55.8) | 30 (50) |
| Age, years | 40.2 ± 7.8 | 41.7 ± 10.5 | 38.6 ± 9.1 | 44.9 ± 7.3 |
| Clinical parameters | ||||
| BMI, kg/m² | 25.0 ± 7.1 | 27.9 ± 4.2 | 23.5 ± 5.1 | 27.6 ± 4.8 |
| Max. CRP, mg/L | 42.0 ± 7.8 | 4.6 ± 2.6 | 3.4 ± 3.1 | 5.0 ± 2.2 |
| Duration of oxygen supplementation, days | 5.0 ± 0.8 | 0.0 ± 0.0 | 0.0 ± 0.0 | 0.0 ± 0.0 |
| Corticosteroids, n (%) | 134 (61.5) | 0 (0) | 45 (86.5) | 7 (11.7) |
| Smoking habit | 15 (6.9) | 3 (5) | 2 (3.8) | 3 (5) |
| Occupational hazard | 2 (0.9) | 0 (0) | 1 (1.9) | 1 (1.7) |
| Comorbidities | ||||
| Chronic respiratory diseases, n (%) | 0 (0) | 0 (0) | 0 (0) | 0 (0) |
| Diabetes mellitus, n (%) | 48 (22.0) | 11 (18.3) | 6 (11.5) | 13 (21.7) |
| Coronary heart disease, n (%) | 32 (14.7) | 9 (15.0) | 7 (13.5) | 14 (23.3) |
| Chronic heart failure, n (%) | 14 (6.4) | 3 (5.0) | 2 (3.8) | 6 (10) |
| Hypertension, n (%) | 75 (34.4) | 19 (31.7) | 19 (36.5) | 27 (45) |
| Chronic kidney disease, n (%) | 9 (4.1) | 2 (3.3) | 0 (0) | 2 (3.3) |
Data are presented as absolute values (%) or mean ± SD. BMI Body mass index, CRP C-reactive protein, COPD Chronic obstructive pulmonary disease
There were differences in the duration of the cough (Fig. 2). In controls, induced cough was longer than in COVID-19 patients (342.5 (277.0; 394.0) vs. 400.5 (359.0; 457.0) ms, p = 0.0001). The duration of the 2nd and 3rd phases was also longer in the non-COVID-19 volunteers (227.5 (190.0; 274.0) vs. 264.0 (203.0; 326.5) ms, p = 0.0095 and 81.0 (61.0; 113.0) vs. 103.5 (72.0; 133.0) ms, p = 0.0046 respectively). At the same time, there were no differences in the duration of the 1 st phase of coughing.
Fig. 2.
Comparison of cough duration and its individual phases in COVID-19 patients, in controls, in patients with asthma and COPD. On the ordinate axis: T, T(c), T(asthma), T(COPD) – cough duration in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; T1, T1(c), T1(asthma), T1(COPD) - duration of the 1 st phase of cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; T2, T2(c), T2(asthma), T2(COPD) - duration of the 2nd phase of cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; T3, T3(c), T3(asthma), T3(COPD) - duration of the 3rd phase of cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; On the abscissa axis: duration in ms
In addition, there were differences in the values of the Q coefficient between COVID-19 patients and control volunteers (0.310 (0.223; 0.454) vs. 0.453 (0.372; 0.619), p = 0.0001) (Fig. 3). Cough sounds in patients with COVID-19 are dominated by high-frequency energy not only of the entire cough, but of each of the three phases (0.392 (0.261; 0.564) vs. 0.456 (0.329; 0.741), p = 0.0183; 0.203 (0.121; 0.313) vs. 0.295 (0.222; 0.414), p = 0.0001; 0.736 (0.479; 1.174) vs. 1.006 (0.774; 1.211), p = 0.0005 respectively).
Fig. 3.
Comparison of coefficients Q, Q1, Q2, Q3 in COVID-19 patients, in controls, in patients with asthma and COPD. On the ordinate axis: Q, Q(c), Q(asthma), Q(COPD) - the ratio of the energy of low and medium frequencies (60–600 Hz) to the energy of high frequencies (600–6000 Hz) of the cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; Q1, Q1(c), Q1(asthma), Q1(COPD) - the ratio of the energy of low and medium frequencies (60–600 Hz) to the energy of high frequencies (600–6000 Hz) of the 1 st phase of cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; Q2, Q2(c), Q2(asthma), Q2(COPD) - the ratio of the energy of low and medium frequencies (60–600 Hz) to the energy of high frequencies (600–6000 Hz) of the 2nd phase of cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; Q3, Q3(c), Q3(asthma). Q3(COPD) - the ratio of the energy of low and medium frequencies (60–600 Hz) to the energy of high frequencies (600–6000 Hz) of the 3rd phase of cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; On the abscissa axis: numerical indicators of the Q coefficient
The frequency of the maximum cough sound energy in patients of the COVID-19 group was significantly higher than in the control group (463.0 (274.0; 761.0) vs. 347.0 (253.0; 488.0) Hz, p = 0.0013). There were no significant differences in frequencies with maximum sound energy of the 1 st, 2nd and 3rd cough phases separately (Fig. 4). However, there was a tendency to increase for a higher frequency of the maximum sound energy of the 2nd cough phase in COVID-19 patients in comparison with induced cough in the control people.
Fig. 4.
Comparison of the maximum frequency of the cough sound energy in COVID-19 patients, in controls, in patients with asthma and COPD. On the ordinate axis: Fmax, Fmax(c), Fmax(asthma), Fmax(COPD) - the frequency of the maximum sound energy of the cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; Fmax1, Fmax1(c), Fmax1(asthma), Fmax1(COPD) - the frequency of the maximum sound energy of the 1 st phase of the cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; Fmax2, Fmax2(c), Fmax2(asthma), Fmax2(COPD) - the frequency of the maximum sound energy of the 2nd phase of the cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; Fmax3, Fmax3(c), Fmax3(asthma), Fmax3(COPD) - the frequency of the maximum sound energy of the 3rd phase of the cough in the COVID-19 group, in control group, in patients with asthma, in patients with COPD, respectively; On the abscissa axis: numerical indicators of the frequency of the maximum energy of the cough sound in Hz, x 1000
The duration of the cough sound in asthma patients was significantly higher in general (495.5 (420.5; 605.0) vs. 342.5 (277.0; 394.0) ms, p = 0.0001) and in the 2nd phase (363.0 (276.5; 465.5) vs. 227.5 (190.0; 274.0) ms, p = 0.0001) in comparison with the cough duration in COVID-19 patients (Fig. 2). The Q coefficient in asthma patients was significantly lower in general (0.287 (0.172; 0.359) vs. 0.310 (0.223; 0.454), p = 0.0222) and significantly higher in the 1 st phase (0.608 (0.428; 0.687) vs. 0.392 (0.261; 0.564), p = 0.0001), however, no differences were found in the 2nd and 3rd phases (Fig. 3). The frequency of maximum cough sound energy in asthma patients was significantly higher than in COVID-19 patients (1163.0 (373.0; 1736.5) vs. 463.0 (274.0; 761.0) Hz, p = 0.0001) (Fig. 4).
The cough duration in COPD patients was significantly higher in general (442.5 (367.5; 538.0) vs. 342.5 (277.0; 394.0) ms, p = 0.0001) and in the 2nd phase in comparison with the cough duration in COVID-19 patients (303.5 (246.5; 420.0) vs. 227.5 (190.0; 274.0), p = 0.0001) (Fig. 2). The Q coefficient in COPD patients was significantly higher than in COVID-19 patients (0.428 (0.318; 0.546) vs. 0.310 (0.223; 0.454), p = 0.0001) (Fig. 3). The frequency of maximum cough sound energy in patients with COPD was significantly lower in general (371.5 (327.5; 468.5) vs. 463.0 (274.0; 761.0) Hz, p = 0.0126) and in the 2nd phase (414.0 (287.0; 554.5) vs. 851.0 (374.0; 1507.0) Hz, p = 0.0001) than in COVID-19 patients; no differences were found in the 1 st and 3rd phases (Fig. 4).
Discussion
Our study demonstrated that cough in COVID-19 patients and induced cough in control volunteers differ in time-frequency parameters, which may be of some importance in the differential diagnosis of cough. The most significant indicators for diagnosis are those of the second phase of cough sounds, as this phase corresponds to the passage of air through the respiratory tract. Cough in COVID-19 is characterized by a shorter duration and a predominance of higher frequencies, compared with induced cough in control individuals. With regard to Q and Q2, this can be explained by the presence of edema of the lower respiratory tract walls, elements of bronchospasm and/or excessive mucus production. In addition, significant differences were found in some time-frequency parameters of cough sounds in asthma and COPD patients in comparison with COVID-19 patients, based on which the differentiation of COVID-19 and these highly common diseases of the respiratory tract is possible.
Recently, attempts have been made to analyze the various characteristics of cough in patients with various respiratory diseases. Rudraraju et al. described the features of cough sounds in normal condition and in various respiratory diseases, which they combined into groups: obstruction - narrowed (asthma, bronchitis) and dilated (COPD) airways; restriction – interstitial lung disease and fluid-filled (pneumonia). In addition, the study compared the obtained audiograms of cough sounds with spirometry data and the clinical picture. The main goal of the study was to use machine learning to predict the nature of respiratory pathology (obstruction and/or restriction) as an equivalent to spirometry [8].
Umayahara et al. created a mobile application on IOS that allows evaluating the cough peak flow (CPF) by cough sounds. The application takes into account age, sex, height and weight. It is also possible to enter indicators of respiratory function such as forced expiratory volume in one second, vital capacity, forced vital capacity, previously recorded CPF values, as well as the repetitive saliva swallowing test (RSST), which identifies patients with aspiration. Since the Bland-Altman analysis did not reveal a systematic error between the measured and estimated CPFs, this development can be used in everyday clinical practice [9].
Chung et al. proposed the diagnosis of pneumonia using the spectral analysis of cough sounds performed by artificial intelligence. The accuracy of this algorithm (84.9%) was compared with the diagnosis made by pulmonologists based on their own hearing (56.4%), and they did not know other data about patients [10].
Currently research is also being conducted to explore the possible application of cough sounds analysis in the diagnosis and evaluation of the treatment effectiveness of the respiratory diseases. So, Toop et al. attempted to create a portable system that would allow diagnosing asthma based on the analysis of cough sounds. The system recorded the sounds of coughing in a free sound field by using a microphone placed in the room with a patient, as well as by using a contact microphone attached to the chest. There was also a device for recording airflow from the oral cavity during coughing. Initially the system was used to study the effect of exercise on the sound of coughing in asthma. Unfortunately, these scientific developments have not received practical application [11].
Importantly, there is data from other authors on the study of cough in patients with COVID-19. Mouawad and colleagues used a symbolic nonlinear dynamics analysis to assess and automatically diagnose COVID-19 based on cough recordings and pronunciation of vowels “ah”, “oh”, “eh”. The authors used recurrence plots (RP) with a sufficient number of dots, and then conducted recurrence quantification analysis (RQA). The method demonstrated high diagnostic accuracy with the use of cough sounds and the pronunciation of the aforementioned vowels (97% and 99%, respectively) [14].
Furthermore, Nguyen et al. diagnosed COVID-19 by Log-Mel spectrograms of cough sound recordings based on short-time Fourier transform and logarithmic calculation. In contrast to our study, visual distinctive features of the spectral characteristics of cough were revealed. The authors found that the features of the spectrograms of COVID-19 negative coughs are more clear and sparse, in contrast to COVID-19-positive ones - they are more dense and noisy. Neural networks were used to extract features. The proposed vision-based framework can be built into mobile applications to detect COVID-19 from home [15].
Interestingly, Sharma and colleagues presented an audio texture analysis of the sounds of coughing, breathing and speaking in their study. Texture analysis of cough and breath sounds was performed for the following 5 classes: COVID-19 positive with cough, COVID-19 positive without cough, healthy people with cough, healthy people without cough, and asthmatic cough. There were only two classes for speech sounds: COVID-19 positive and COVID-19 negative. The spectral analysis of sounds was also taken as a basis, the frequency-time parameters were presented in the form of spectrograms. Visually, it was noted that COVID-19 cough is more consistent in nature, has a noise-like texture that is uniform across all frequency bands, while COVID-19 negative cough has a low texture and is concentrated within a few frequency bands. To analyze the textural characteristics of the image, local binary patterns (LBP) and the Haralick’s features were used [16].
Based on the spectral analysis of cough sounds using neural networks, Melek Manshouri et al. diagnosed the COVID-19. The authors concluded that cough in COVID-19 can be distinguished from cough in other diseases using feature extraction and classification techniques. As an effective feature extraction method, they chose the spectral analysis of cough sounds based on the Fourier transform and the Mel-frequency cepstral coefficients. Of the classification methods, the support vector algorithm was applied to the processed signals. The sensitivity and specificity of the described technique in the diagnosis of cough in COVID-19 was 98.6% and 91.7%, respectively [17]. However, it is important to understand that the use of neural networks, even with high-quality training and obtaining results with high accuracy, still does not allow obtaining the main thing - specific parameters and characteristics, on the basis of which cough could be differentiated.
The relatively small sample size of our study can be considered as a limitation. In addition, we analyzed individual cough sounds rather than cough cascades. Changes in cough sounds in patients with the first symptoms of COVID-19, during the development of the disease and recovery also need to be studied. In addition, the set of cough parameters was limited and could be expanded.
Conclusions
Our study aimed at improving the diagnosis of COVID-19 using the method of cough sounds analysis. Significant differences in individual parameters of cough sounds (cough duration, the ratio of low and medium frequency energy (60–600 Hz) to high frequency energy (600–6000 Hz) and the frequency of maximum sound energy) allowed us to differentiate cough in the context of COVID-19. Promising areas for continuing the study are the identification of differences or the absence of such differences in the analysis of individual cough sounds in cough cascades and the expansion of the set of studied sound parameters.
Analysis of cough sounds is a promising diagnostic method that can be used to diagnose and monitor treatment not only for COVID-19, but also for other diseases presenting with cough. Due to analysis, the time-frequency characteristics of the cough sound are evaluated, which may have their own characteristics in various diseases. This method will open up new opportunities in the diagnosis of diseases manifested by coughing, and in an objective assessment of the treatment effectiveness based on an individual approach.
Authors’ contributions
“Research concept and design: AVB and SNA. Statistical processing: ESO, ONC and GGP. Collecting and processing of data: AVP, SNF, AVM, SAK, RET, AVBA and VRD. Manuscript drafting and writing: AVB and SNA. Manuscript writing and editing: DK, ESO, ONC, AVP, SNF, TAC, AVM, GGP, SAK, RET, AVBA and VRD. All the authors made a significant contribution to the literature search and analytical work for the preparation of the article. All authors read and approved the final version before publication.”
Funding
There is no funding related to this study.
Data availability
“Data are available from the corresponding author upon the reasonable request.”
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the principles of the Helsinki Declaration of the World Medical Association and it was approved by the local ethics committee “Ethics Committee of Voronezh State Medical University named after N. N. Burdenko, Ministry of Health of Russian Federation” (approval number: Protocol No. 7, September 18th, 2021). Written informed consent was obtained from each patient to participate in the study.
Consent for publication
n/a.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Guan WJ, Ni ZY, Hu Y, et al. Clinical characteristics of coronavirus disease 2019 in China. N Engl J Med. 2020;382:1708–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chams N, Chams S, Badran R, et al. COVID-19: A multidisciplinary review. Front Public Health. 2020;8:383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ovsyannikov ES, Avdeev SN, Budnevskiy AV, Drobysheva ES, Kravchenko AY. COVID-19 and chronic obstructive pulmonary disease: what is known about the unknown. Tuberculosis Lung Dis. 2021;99:6–15. [Google Scholar]
- 4.Ovsyannikov ES, Аvdeev SN, Budnevskiy AV, Drobysheva ES, Savushkina IA. Bronchial asthma and COVID-19: comorbidity issues. Tuberculosis Lung Dis. 2021;99:6–14. [Google Scholar]
- 5.Berekaa MM. Insights into the COVID-19 pandemic: origin, pathogenesis, diagnosis, and therapeutic interventions. Front Biosci (Elite). 2021;13:117–39. [DOI] [PubMed] [Google Scholar]
- 6.Anka AU, Tahir MI, Abubakar SD, et al. Coronavirus disease 2019 (COVID-19): an overview of the immunopathology, serological diagnosis and management. Scand J Immunol. 2021;93:e12998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Song WJ, Hui CKM, Hull JH, et al. Confronting COVID-19-associated cough and the post-COVID syndrome: role of viral neurotropism, neuroinflammation, and neuroimmune responses. Lancet Respir Med. 2021;9:533–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Rudraraju G, Palreddy S, Mamidgi B, et al. Cough sound analysis and objective correlation with spirometry and clinical diagnosis. Inf Med Unlocked. 2020;19:100319. [Google Scholar]
- 9.Umayahara Y, Soh Z, Sekikawa K, Kawae T, Otsuka A, Tsuji TA. Mobile cough strength evaluation device using cough sounds. Sens (Basel). 2018;18:3810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chung Y, Jin J, Jo HI, et al. Diagnosis of pneumonia by cough sounds analyzed with statistical features and AI. Sens (Basel). 2021;21:7036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Toop LJ, Dawson KP, Thorpe CW. A portable system for the spectral analysis of cough sounds in asthma. J Asthma. 1990;27:393–7. [DOI] [PubMed] [Google Scholar]
- 12.Budnevsky AV, Avdeev SN, Ovsyannikov ES, et al. Spectral analysis of cough sounds of patients with COVID-19. Pulmonologiya. 2022;32:834–41. [Google Scholar]
- 13.Morice AH, Kitt MM, Ford AP, et al. The effect of gefapixant, a P2X3 antagonist, on cough reflex sensitivity: a randomised placebo-controlled study. Eur Respir J. 2019;54:1900439. [DOI] [PubMed] [Google Scholar]
- 14.Mouawad P, Dubnov T, Dubnov S. Robust detection of COVID-19 in cough sounds: using recurrence dynamics and variable Markov model. SN Comput Sci. 2021;2:34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Nguyen LH, Pham NT, Do VH, et al. Fruit-CoV: an efficient vision-based framework for Speedy detection and diagnosis of SARS-CoV-2 infections through recorded cough sounds. Expert Syst Appl. 2023;213:119212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sharma G, Umapathy K, Krishnan S. Audio texture analysis of COVID-19 cough, breath, and speech sounds. Biomed Signal Process Control. 2022;76:103703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Melek Manshouri N. Identifying COVID-19 by using spectral analysis of cough recordings: a distinctive classification study. Cogn Neurodyn. 2022;16:239–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
“Data are available from the corresponding author upon the reasonable request.”