New cough science publications vetted and collected in one place
How should automatic cough detection algorithms be evaluated?
Key Takeaway: Over 60% of studies of cough counting algorithms report specificity and accuracy, yet these are unhelpful for assessing algorithm performance as they are highly sensitive to class imbalance in the data set. Instead, specificity and false positive rate are recommended.
Why It Matters: This paper is an important step in ensuring that the research community has a clear and shared definition of the relevant metrics for assessing the performance of cough counting algorithms. Widespread usage of objective longitudinal cough counting will happen faster, if algorithm performance is widely understood and can be compared.
What causes cough in pulmonary fibrosis and how should we treat it?
Key Takeaway: Cough is a frequently occurring symptom for patients with pulmonary fibrosis, it may precede diagnosis, and besides dyspnoea it is one of the most distressing symptoms - yet the underlying mechanisms remain incompletely understood. Regarding treatment, management of inflammation in ILD, novel antitussives aimed at peripheral and central nerves, as well as non-pharmacological measures such as cough suppression speech and language therapy, should all be explored.
Why It Matters: The paper summarizes many areas of potentially fruitful research related to ILD/pulmonary fibrosis and highlights the critical importance of ensuring studies have reliable, objective data. Understanding the etiology of cough in patients with ILD is crucial for developing therapies that effectively reduce cough and improve quality of life.
How might asthma be detected using cough sounds?
Key Takeaways: A machine-learning model designed to detect asthma from cough sounds achieved accuracy rates of 98.24% and 96.91% using tenfold and leave-one-subject-out cross-validation strategies, respectively. The study developed and utilized a new cough sound database of over 1,000 patients and introduced a novel cough sound classification architecture.
Why it matters: The study lays a promising foundation for developing an efficient and reliable screening tool that is accessible and easy to implement, being computationally lightweight. Such a tool could address under-diagnosis and under-treatment, particularly in low- and middle-income countries, and facilitate earlier detection of asthma.
Plus…
A Cough Science Forum Special
Watch Hyfe’s Cough Science Forum, a quarterly Zoom event hosted by Hyfe that features recent and ongoing research in cough science, available via on-demand video for the first time.
Agenda:
Cough monitoring as part of remote home monitoring program for people with COPD after a recent exacerbation (using an earlier version of Hyfe’s technology)
Speaker: Jansen Zhou (Robert Wu), University Health Network, Sinai Health System, University of Toronto
Chronic Bronchitis and Cough: A 5-Patient Snapshot
Speaker: Dan Copeland, Renovion
Use of a digital device to detect cough in heart failure patients
Speaker: Matthew Lassman, McGill University Health Center
Accelerating Cough-Based Tuberculosis Detection Through Open Collaboration: The CODA TB DREAM Challenge
Speaker: Stephen Burkot, Global Health Labs
Watch the full video here.
Reminder: World Bronchiectasis Conference - The leading conference for bronchiectasis science will be in Dundee, July 4-6th, exploring the latest in research and treatment. Hyfe will present emerging data on continuously monitoring cough to assess bronchiectasis treatment response.
The 13th London International Cough Symposium - Imperial College London will host leading cough scientists and clinicians on 18th-19th July for a review of the latest clinical and research advances, evolving ideas and new approaches to the management of chronic cough. Hyfe will be there, please reach out if you’d like to meet.
