COPD

Australian smartphone app for cough detects worsening COPD symptoms


Australian respiratory clinicians are turning to AI and smartphones to pick up cough sounds of acute COPD exacerbation (AECOPD), prompting early intervention that could  prevent hospitalisations.

Using technology similar to that used in speech recognition software, respiratory specialists and researchers from Queensland and Western Australia have developed a smartphone-based set of algorithms that can be ‘trained’ to diagnose and differentiate between common respiratory diseases, like isolated upper respiratory disease, lower respiratory disease, COPD, COPD exacerbations and acute asthma, seen in adults.

Matching signatures from a database of more than 1,200 cough recordings from people with known clinical diagnoses, the algorithm incorporates patient-reported features: age, fever, and presence of new cough – along with audio data from five coughs from the user – to rapidly detect worsening of symptoms, potentially preventing AECOPD.

According to findings from a study in 86 patients with known COPD, the algorithm correctly identified the presence of AECOPD in 82.6% cases with absence of AECOPD correctly identified in 91.0% of individuals compared to expert clinical assessment using all available investigations and results in patients’ medical records, including the treating team’s discharge diagnosis.

Accuracy in mild COPD

Speaking to the limbic, sleep and respiratory physician and principal study investigator Dr Scott Claxton from Genesis Care in Perth said an important finding was that diagnostic accuracy was maintained in milder cases of AECOPD (PPA: 79.2%, 95% CI: 68.0–87.8%), who typically present to primary care.

Accuracy was also maintained in patients aged greater than 65 where co-morbidities were likely to be greater. Reassuringly, says Dr Claxton, accuracy was maintained in those with co-morbid heart failure – a common co-morbidity with COPD that causes similar symptoms, including exertional breathlessness and nocturnal cough.

Diagnosis of COPD and AECOPD in this group is particularly complicated by the ventilatory defects exhibited by patients with heart failure, which obscure the diagnostic airflow limitation characteristic of COPD, notes Dr Claxton.

Providing virtually instantaneous diagnostic results that are not reliant upon monitoring symptom decline over several days nor upon subjective interpretation of patient symptoms, Dr Claxton believes the algorithm has the potential to improve self management of AECOPD particularly in the setting of COVID and intermittent lockdowns.

“This gives us the opportunity now to perform a relatively simple test to pick up this condition and it’s particularly useful now where face-to-face consultations are difficult or being discouraged because of COVID. It’s utility is to give some sort of strength to a diagnosis on Telehealth consultation or even in remote locations where there is not a lot of medical support available this relatively simple to perform diagnostic test can give you some base for what’s going on.”

Screening tool

While the algorithm doesn’t replace existing lung function tests like spirometry for a COPD diagnosis, Dr Claxton said the app could be used as a screening tool and by patients to support written action plans. While these are used to guide patients to initiate therapy with oral steroids and antibiotics without direct clinician involvement Dr Claxton says the requirement of the patient to self-recognise their  AECOPD is a task which many patients may be unable to perform.

Around two-thirds of patients cannot recognise that the worsening of at least one of four key symptoms – dyspnoea, sputum amount, and colour – represents an exacerbation of their COPD, are confused over the use of the term exacerbation, and misinterpret the presence and severity of their AECOPD, according to the study.

“We’ve got some data to suggest that this is a useful screening tool – that once COPD is picked up we need to proceed on to spirometry for confirmation.

“Here in the COPD-diagnosed exacerbation group we’ve now got evidence that the algorithm supports a diagnosis of exacerbation and the need to start steroids and antibiotics so that you’re heading down the right track with confidence as opposed to being something else where steroids are not appropriate.”

“Ultimately it helps avoid over treatment and people coming into the healthcare environment during a pandemic when that in itself is quite risky at the moment.”

The study is published in Digital Medicine.

Already a member?

Login to keep reading.

OR
Email me a login link
logo

© 2022 the limbic