Chronic Obstructive Pulmonary Disease

Biography

Biography: Udantha Abeyratne

Abstract

Introduction: spirometry is an essential tool to diagnose COPD. The reliability of spirometry depends on the expertise of the technologist who administers the test and the patient’s efforts. Laboratory spirometry is not suitable for patient-centred COPD management plans in home settings. Automated technologies independent of patient efforts to monitor COPD are urgently needed for laboratory as well as home use. Cough is a cardinal symptom of COPD but it is not sufficiently used in diagnosing COPD beyond noting it as a symptom. Adventitious sounds such as crackles and wheezing generated in a diseased lung as well as information on airflow limitations due to obstructive illnesses should be available via a cough. Our target is to develop mathematical techniques for cough analysis for the diagnosis of COPD and associated illnesses such as asthma and emphysema. We also implement this technology on a smartphone, together with automated cough detection and counting software. No network connection or extra attachements are required. Method: cough sounds were recorded at the Princess Alexandra Hospital (PAH), Brisbane from adults referred for routine spirometry. We recorded voluntary cough events using an iPhone (in air, outside the mouth) and analyzed them following a technique similar to our pediatric pneumonia technology (Ann Biomed Eng, 41(11):2448-62)). We explored the performance of the method using a leave-one-out validation technology, against the clinical diagnosis supported by spirometry and other tests deemed clinically necessary. At the time of writing, our database at PAH consists of 66 subjects (COPD (n=15), no respiratory disease (n=42) and obesity hypoventilation (n=6)). Results: the cough sound based algorithm achieved a classification sensitivity and specificity of 80% and 88% respectively in separating COPD patients from the rest. The technology can be implemented on a smart phone such as an iPhone. Conclusion: our preliminary results were obtained on a small dataset, but they suggest cough sounds carry vital information on COPD. Results we have obtained – accuracies in the order of 90% – from pediatric subjects on illnesses such as pneumonia, bronchiolitis and asthma/viral-wheeze further encourage us to apply our methods on adult populations. Our database of adult patients is growing and we expect to present further improvements to the technology at the conference.