Background and Objective:We live in the era of precision health in which parameters of clinical importance are quantified and used to tailor therapies. However, cough is a common and informative sign and symptom which is generally not quantified. Recently, advances in acoustic artificial intelligence (AI) have enabled accurate, passive and privacy preserving continuous cough monitoring (CCM). Analytically and clinically validated CCM provides a reproducible and potentially clinically useful measurement of cough as a physiological signal-quantifying the burden and day-to-day variability, supporting earlier detection of deterioration, offering a sensitive endpoint in clinical trials and even contributing to population syndromic surveillance. The objective of this review is to critically appraise the evidence for such use cases and identify future directions for research in this field and clinical adoption. Methods:In this article, we reviewed the literature and summarised insights gained from CCM across various diseases, and discussed future directions for this emerging field. Key Content and Findings:CCM has already provided novel and important insights into the biology and therapy of specific diseases such as refractory and unexplained chronic cough, chronic obstructive pulmonary disease (COPD), bronchiectasis, congestive heart failure and gastroesophageal reflux. In addition, it has demonstrated aspects of cough as individuals go about their daily lives, such as the inter and intra-subject variability in daily cough frequency, diurnal and episodic patterns of cough, and the correlation between its subjective and objective measurement. Predictions are presented about future research and uses of cough monitoring. Conclusions:AI-enabled CCM is a powerful new tool that is already improving cough research and patient care.

RATIONALE: Periods of intense coughing (bouts, bursts, fits, epochs) are particularly problematic for coughers but not well reflected in simple tabulations of daily cough frequency. Traditional definitions of cough bouts rely on arbitrary parameters: fixed inter-cough intervals and minimum cough counts, which may not accurately characterize the true pattern and impact of cough. This study examines the effects of varying bout definitions on cough pattern assessment using continuous cough monitoring. METHODS: Two chronic coughers’ data, collected via continuous cough monitoring over two weeks with time stamped cough events, were analyzed to understand how different definitions of cough bouts influence measurement outcomes. Bout definitions were varied using inter-cough intervals of 2, 4, and 6 seconds and minimum cough counts of 2, 4, and 6. Daily bout counts, coughs per bout, and bout duration were examined for each parameter set to assess how these variations reflect different cough patterns. RESULTS: Varying the inter-cough interval and minimum cough count significantly altered the resulting metrics for cough bout counts, duration, and frequency. Short inter-cough intervals and low cough count thresholds inflated bout frequency, particularly for the participant coughing frequently but briefly. In contrast, longer inter-cough intervals and higher minimum cough counts revealed more sustained coughing episodes, emphasizing the diverse coughing patterns – and potentially the severity of cough experienced – between individuals. In panel A, the dashed lines indicate the average daily cough count over the two-week period: 1,420 coughs/d for participant-1 and 691 coughs/d for participant-2. Panels B and C highlight that participant 1 primarily coughs twice in rapid succession, while participant 2 tends to cough in longer-duration waves, suggesting the possibility that the cough experience may be more severe for participant-2. CONCLUSIONS: Measuring periods of intense coughing may better capture the patient's experience and improve the correlation between objective and subjective cough measurement. Arbitrary cough bout definitions can misrepresent cough severity by either overestimating or underestimating coughing patterns, depending on the selected parameters. Continuous cough monitoring offers a nuanced understanding of cough dynamics and may support refined bout metrics that more accurately reflect clinical needs. Further research may establish standardized, evidence-based parameters for cough bout definitions that account for individual coughing patterns and patient experience, enhancing their application in clinical studies and therapeutic evaluations. The frequency and pattern of bout/count discrepancies observed in these two patients will be explored in a cohort of patients, and will be presented at the event.

THE EUROPEAN Respiratory Society (ERS) Congress 2025 showcased continuous cough monitoring across sensor modalities, a pragmatic 7-day monitoring standard, and early signals for efficacy and tolerability in real studies, positioning cough as a scalable biomarker and clinical endpoint for respiratory care.

RATIONALE: Accurate assessment of cough frequency is essential for evaluating cough as a symptom, as a disease indicator and as a biomarker in the development of novel therapeutics. Traditional monitoring durations (e.g., 24 hours) fail to capture the variability in cough frequency observed in real-world conditions, potentially leading to skewed results in clinical studies. This study evaluates the optimal monitoring duration to achieve reliable estimates of cough frequency, proposing a 7-day monitoring standard based on statistical modeling of continuous cough data. METHODS: Using real-world cough data from over 30 individuals with problematic cough, monitored for at least 20 hours per day over 60 days, we analyzed daily and hourly cough frequency distributions. The data revealed that daily cough counts do not fit conventional statistical distributions, complicating analysis. However, hourly cough counts showed a robust fit to zero-inflated negative binomial (ZINB) distributions, enabling more reliable modeling. We used ZINB-based simulations to model 504 hours (21 days) of cough data across various frequency profiles, calculating the minimum monitoring duration needed to achieve consistent estimates of hourly cough frequency within 20% of true values. RESULTS: Simulations demonstrated that most individuals, especially those with a coefficient of variation (CV) between 0.8 and 1.65, reached stable cough frequency estimates within a 7-day (168 hours) monitoring period (Figure 1). Subjects with lower variability (CV < 1) required fewer days, while those with higher variability (CV > 1.65) needed closer to 8 days for reliable data. Across all profiles, 7 days was sufficient to capture accurate cough frequency for 83% of participants, suggesting it as a potentially pragmatic balance between scientific rigor and operational feasibility. CONCLUSION: A 7-day monitoring duration provides a robust, standardized timeframe for assessing cough frequency in clinical studies, offering a significant improvement over the traditional 24-hour period. This duration captures inter- and intra-individual cough variability, enhancing the accuracy of cough-related endpoints in research. While future work may explore tailored monitoring durations for specific populations, a 7-day period is recommended as the default for most studies involving cough frequency assessment.

Background:Open-access data challenges can accelerate innovation in artificial intelligence-based tools. In the Cough Diagnostic Algorithm for Tuberculosis (CODA TB) DREAM Challenge, we developed and independently validated cough sound-based artificial intelligence algorithms for tuberculosis screening. Methods:We included data from 2143 adults with ≥2 weeks of cough from outpatient clinics in India, Madagascar, the Philippines, South Africa, Tanzania, Uganda, and Vietnam. A standard tuberculosis evaluation was completed, and ≥3 solicited coughs were recorded using a smartphone. We invited teams to develop models using training data to classify microbiologically confirmed tuberculosis disease using (1) cough sound features only and/or (2) cough sound features with routinely available clinical data. After 4 months, they submitted the algorithms for independent test set validation. Models were ranked by area under the receiver operating characteristic curve (AUROC) and partial AUROC (pAUROC) to achieve at least 80% sensitivity and 60% specificity. Results:Eleven cough models and 6 cough-plus-clinical models were submitted. AUROCs for cough models ranged from 0.69 to 0.74, and the highest performing model achieved 55.5% specificity (95% confidence interval, 47.7%-64.2%) at 80% sensitivity. The addition of clinical data improved AUROCs (range, 0.78-0.83); 5 of the 6 models reached the target pAUROC, and the highest performing model had 73.8% specificity (95% confidence interval, 60.8%-80.0%) at 80% sensitivity. The AUROC varied by country and was higher among male and human immunodeficiency virus-negative individuals. Conclusions:In a short period, an open-access data challenge facilitated the development of new cough-based tuberculosis algorithms and demonstrated potential as a tuberculosis screening tool.

RATIONALE: Chronic cough, which persists despite guideline based treatment, is refractory chronic cough (RCC). Given the lack of treatment options available for patients with RCC and the evidence supporting cognitive-behavioral techniques in overriding cough reflex, we describe a proof of concept scientifically driven therapeutic solution, delivered digitally, based on cough monitoring and targeted cough suppression. CoughPro, a globally available consumer wellness smartphone app, enables continuous cough monitoring and incorporates a cough management feature based on evidence-based behavioral cough suppression techniques. METHODS: The data from consenting users with problematic cough, who monitored their cough for at least 7 days before and at least 7 days after the delivery of the cough management techniques inside the app, were included in this preliminary analysis. We analyzed the daily cough rate (coughs/hour) and the daily number of cough bursts (defined as a cluster of 4 or more coughs, none of which is separated by 5 seconds or more) before and after the introduction of cough management techniques. RESULTS: In this proof-of-concept study, a cohort of 14 chronic cough patients with an average baseline cough rate of 11.5 coughs/hr and 23 cough bursts per day was analyzed, demonstrating promising results in cough suppression and control of cough frequency and bursts. A 31% reduction in daily cough rate (Fig. A) and a 52% reduction in the total number of cough bursts per day (Fig. B) were observed following the introduction of the digital behavioral cough management feature. No adverse events were reported. CONCLUSION: This novel proof of concept therapeutic solution for RCC suggests that digitally delivered behavioral cough suppression techniques with objective cough monitoring data might provide motivation, adherence and relief in refractory chronic cough patients. This hypothesis for a novel digital therapeutic solution should be evaluated in a definitive, sham-controlled clinical trial.

The ability to passively and continuously monitor coughing would significantly improve cough management and research. To date there is no automated clinically validated cough monitor that can be routinely used in clinical care and research. Here we describe the validation of such an automated cough monitor. To assess the overall performance of the Hyfe Cough Monitoring System when used by individuals with problematic cough, under common living conditions This multicenter observational study compared the results of the Hyfe CoughMonitor wrist-worn device with manually counted coughs in subjects with a variety of etiologies as they went about their usual daily activities. We collected 24 hours of continuous sounds from subjects while they simultaneously wore a CoughMonitor and an audio recorder. Coughs were labelled by multiple trained annotators who listened to the continuous audio recordings using validated methodology. The time stamps of these human-detected coughs were compared to those of the CoughMonitor to determine the system’s overall performance using event-to-event and hourly rate correlation analyses. Over the 546 hours monitored, 4454 cough events were recorded; The overall sensitivity was 90.4% (95% CI of 88.3% to 92.2%). The overall false positive rate was 1.03 false positives per hour (95% CI of 0.84 to 1.24). The overall correlation between manual and CoughMonitor measured hourly coughing was high (Pearson correlation coefficient of 0.99 with OLS slope 0.94 and OLS intercept 0.68). The present analysis of cough events demonstrated that the Hyfe CoughMonitor accurately reflects them with a high sensitivity and a low false positive rate. Future studies should confirm its potential role in the management of patients with cough in clinical practice. Clinicaltrials.gov: NCT05723159

BackgroundTreatment of chronic cough remains a challenge. We hypothesised that inhaled alkaline hypertonic divalent salts (alkaline HDS) might provide relief for refractory chronic cough by laryngeal and tracheal hydration.MethodsWe conducted an exploratory, single-blinded, nasal saline-controlled study in 12 refractory chronic cough patients to examine cough suppression efficacy of an alkaline HDS composition (SC001) at pH 8 or pH 9 administered by nasal inhalation. As control, we used nasal saline with the same hand-held pump spray aerosol device. Each subject was monitored continuously using a digital cough monitor watch for 1 week of baseline, 1 week of control treatment and 1 week of active treatment.ResultsBaseline daily cough rates ranged from 4 to 34 coughs·h−1with mean visual analogue score 65±17 pre- and post-baseline testing. Control-adjusted efficacy of cough rate reduction ranged from 15% (p=0.015) (from Day 1) to 23% (p=0.002) (from Day 3). Control-adjusted efficacy was highest with SC001 pH 9 (n=5), ranging from 25% (p=0.03) (from Day 1) to 35% (p=0.02) (from Day 3), and lowest for SC001 pH 8 (n=7), ranging from 9% (p=0.08) (from Day 1) to 16% (p=0.02) (from Day 3). Hourly cough counts and visual analogue score correlated for baseline (r=0.254, p=0.02) and control (r=0.299, p=0.007) monitoring weeks. Treatment improved this correlation (r=0.434, p=0.00006). No adverse events were reported.ConclusionsAlkaline (pH 9) HDS aerosol is a promising treatment for refractory chronic cough and should be further evaluated.

ABSTRACTCough is a common and commonly ignored symptom of lung disease. Cough is often perceived as difficult to quantify, frequently self-limiting, and non-specific. However, cough has a central role in the clinical detection of many lung diseases including tuberculosis (TB), which remains the leading infectious disease killer worldwide. TB screening currently relies on self-reported cough which fails to meet the World Health Organization (WHO) accuracy targets for a TB triage test. Artificial intelligence (AI) models based on cough sound have been developed for several respiratory conditions, with limited work being done in TB. To support the development of an accurate, point-of-care cough-based triage tool for TB, we have compiled a large multi-country database of cough sounds from individuals being evaluated for TB. The dataset includes more than 700,000 cough sounds from 2,143 individuals with detailed demographic, clinical and microbiologic diagnostic information. We aim to empower researchers in the development of cough sound analysis models to improve TB diagnosis, where innovative approaches are critically needed to end this long-standing pandemic.

Importance Open-access data challenges have the potential to accelerate innovation in artificial-intelligence (AI)-based tools for global health. A specimen-free rapid triage method for TB is a global health priority. Objective To develop and validate cough sound-based AI algorithms for tuberculosis (TB) through the Cough Diagnostic Algorithm for Tuberculosis (CODA TB) DREAM challenge. Design In this diagnostic study, participating teams were provided cough-sound and clinical and demographic data. They were asked to develop AI models over a four-month period, and then submit the algorithms for independent validation. Setting Data was collected using smartphones from outpatient clinics in India, Madagascar, the Philippines, South Africa, Tanzania, Uganda, and Vietnam. Participants We included data from 2,143 adults who were consecutively enrolled with at least two weeks of cough. Data were randomly split evenly into training and test partitions. Exposures Standard TB evaluation was completed, including Xpert MTB/RIF Ultra and culture. At least three solicited coughs were recorded using the Hyfe Research app. Main Outcomes and Measures We invited teams to develop models using 1) cough sound features only and/or 2) cough sound features with routinely available clinical data to classify microbiologically confirmed TB disease. Models were ranked by area under the receiver operating characteristic curve (AUROC) and partial AUROC (pAUROC) to achieve at least 80% sensitivity and 60% specificity. Results Eleven cough models were submitted, as well as six cough-plus-clinical models. AUROCs for cough models ranged from 0.69-0.74, and the highest performing model achieved 55.5% specificity (95% CI 47.7-64.2) at 80% sensitivity. The addition of clinical data improved AUROCs (range 0.78-0.83), five of the six submitted models reached the target pAUROC, and highest performing model had 73.8% (95% CI 60.8-80.0) specificity at 80% sensitivity. In post-challenge subgroup analyses, AUROCs varied by country, and was higher among males and HIV-negative individuals. The probability of TB classification correlated with Xpert Ultra semi-quantitative levels. Conclusions and Relevance In a short period, new and independently validated cough-based TB algorithms were developed through an open-source and transparent process. Open-access data challenges can rapidly advance and improve AI-based tools for global health. Question Can an open-access data challenge support the rapid development of cough-based artificial intelligence (AI) algorithms to screen for tuberculosis (TB)? Findings In this diagnostic study, teams were provided well-characterized cough sound data from seven countries, and developed and submitted AI models for independent validation. Multiple models that combined clinical and cough data achieved the target accuracy of at least 80% sensitivity and 60% specificity to classify microbiologically-confirmed TB. Meaning Cough-based AI models have promise to support point-of-care TB screening, and open-access data challenges can accelerate the development of AI-based tools for global health. ### Competing Interest Statement PMS is employed by Hyfe AI. ### Funding Statement The CODA TB DREAM Challenge and post-challenge evaluation was funded in part by the Bill & Melinda Gates Foundation. R2D2 was funded by the U.S. National Institutes of Health (U01 AI152087), and the Digital Cough Monitoring study was funded by the Patrick J. McGovern Foundation. SGL is supported by a Junior 1 Salary Award from the Fonds de Recherche Sante Quebec. DJ is supported by funding by the National Institutes of Health. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: Ethical approvals for the studies were obtained from institutional review boards (IRB) in the US (R2D2 TB Network, University of California, San Francisco) and Canada (Digital Cough Monitoring Project, University of Montreal), as well as IRBs in each country in which participants were enrolled. In Vietnam, approval was obtained from the Ministry of Health Ethical Committee for National Biological Medical Research (94/CN-HĐĐĐ), the National Lung Hospital Ethical Committee for Biological Medical Research (566/2020/NCKH) and the Hanoi Department of Health, Hanoi Lung Hospital Science and Technology Initiative Committee (22/BVPHN). In India, approval was obtained from Christian Medical College IRB (13256). In South Africa, approval was obtained from Stellenbosch University Health Research Ethics Committee (17047). In Uganda, approval was obtained from Makerere University, College of Health Sciences, School of Medicine, Research Ethics Committee (2020-182). In the Philippines, approval was obtained from De La Salle Health Sciences Institute Independent Ethics Committee (2020-33-02-A). In Madagascar, approval was obtained from the Comite d Ethique a la Recheche Biomedicale (IORG0000851). In Tanzania, approval was obtained from the Ifakarah Health Institute IRB (31-2021) and the National Institute for Medical Research (NIMR/HQ/R.8a/Vol IX/3805). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The challenge training data and links to the code and write-ups for the model submissions are available at [www.synapse.org/TBcough][1]. Additionally, users can register to submit models for evaluation against the validation data in an ongoing manner. <https://www.synapse.org/TBcough> [1]: http://www.synapse.org/TBcough

Chest X-ray is a commonly used tool during triage, diagnosis and management of respiratory diseases. In resource-constricted settings, optimizing this resource can lead to valuable cost savings for the health care system and the patients as well as to and improvement in consult time. We used prospectively-collected data from 137 patients referred for chest X-ray at the Christian Medical Center and Hospital (CMCH) in Purnia, Bihar, India. Each patient provided at least five coughs while awaiting radiography. Collected cough sounds were analyzed using acoustic AI methods. Cross-validation was done on temporal and spectral features on the cough sounds of each patient. Features were summarized using standard statistical approaches. Three models were developed, tested and compared in their capacity to predict an abnormal result in the chest X-ray. All three methods yielded models that could discriminate to some extent between normal and abnormal with the logistic regression performing best with an area under the receiver operating characteristic curves ranging from 0.7 to 0.78. Despite limitations and its relatively small sample size, this study shows that AI-enabled algorithms can use cough sounds to predict which individuals presenting for chest radiographic examination will have a normal or abnormal results. These results call for expanding this research given the potential optimization of limited health care resources in low- and middle-income countries.

Cough is a common and commonly ignored symptom of lung disease. Cough is often perceived as difficult to quantify, frequently self-limiting, and non-specific. However, cough has a central role in the clinical detection of many lung diseases including tuberculosis (TB), which remains the leading infectious disease killer worldwide. TB screening currently relies on self-reported cough which fails to meet the World Health Organization (WHO) accuracy targets for a TB triage test. Artificial intelligence (AI) models based on cough sound have been developed for several respiratory conditions, with limited work being done in TB. To support the development of an accurate, point-of-care cough-based triage tool for TB, we have compiled a large multi-country database of cough sounds from individuals being evaluated for TB. The dataset includes more than 700,000 cough sounds from 2,143 individuals with detailed demographic, clinical and microbiologic diagnostic information. We aim to empower researchers in the development of cough sound analysis models to improve TB diagnosis, where innovative approaches are critically needed to end this long-standing pandemic.

PurposeWe determined the cough counts and their variability in subjects with persistent cough for 30 days.MethodsThe Hyfe cough tracker app uses the mobile phone microphone to monitor sounds and recognizes cough with artificial intelligence-enabled algorithms. We analyzed the daily cough counts including the daily predictability rates of 97 individuals who monitored their coughs over 30 days and had a daily cough rate of at least 5 coughs per hour.ResultsThe mean (median) daily cough rates varied from 6.5 to 182 (6.2 to 160) coughs per hour, with standard deviations (interquartile ranges) varying from 0.99 to 124 (1.30 to 207) coughs per hour among all subjects. There was a positive association between cough rate and variability, as subjects with higher mean cough rates (OLS) have larger standard deviations. The accuracy of any given day for predicting all 30 days is the One Day Predictability for that day, defined as the percentage of days when cough frequencies fall within that day's 95% confidence interval. Overall Predictability was the mean of the 30-One Day Predictability percentages and ranged from 95% (best predictability) to 30% (least predictability).ConclusionThere is substantial within-day and day-to-day variability for each subject with persistent cough recorded over 30 days. If confirmed in future studies, the clinical significance and the impact on the use of cough counts as a primary end-point of cough interventions of this variability need to be assessed.

Tuberculosis (TB) is an infectious disease caused by the bacterium Mycobacterium tuberculosis and primarily affects the lungs, as well as other body parts. TB is spread through the air when an infected person coughs, sneezes, or talks. Medical doctors diagnose TB in patients via clinical examinations and specialized tests. However, coughing is a common symptom of respiratory diseases such as TB. Literature suggests that cough sounds coming from different respiratory diseases can be distinguished by both medical doctors and computer algorithms. Therefore, cough recordings associated with patients with and without TB seems to be a reasonable avenue of investigation. In this work, we utilize a very large dataset of TB and non-TB cough audio recordings obtained from the south-east of Africa, India, and the south-east of Asia using a fully automated phone-based application (Hyfe), without manual annotation. We fit statistical classifiers based on spectral and time domain features with and without clinical metadata. A stratified grouped cross-validation approach shows that an average Area Under Curve (AUC) of approximately 0.70 $\pm$ 0.05 both for a cough-level and a participant-level classification can be achieved using cough sounds alone. The addition of demographic and clinical factors increases performance, resulting in an average AUC of approximately 0.81 $\pm$ 0.05. Our results suggest mobile phone-based applications that integrate clinical symptoms and cough sound analysis could help community health workers and, most importantly, health service programs to improve TB case-finding efforts while reducing costs, which could substantially improve public health.