An algorithm influenced by cough sound recognition and symptom reporting conducted using a smartphone was found to be effective at diagnosing patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD).
A smartphone-based algorithm that uses cough sound recognition and symptom documentation from patients was found to be effective at diagnosing acute exacerbations of chronic obstructive pulmonary disease (AECOPD), according to a recent study published in npj Digital Medicine.
“The algorithm has the potential to improve the diagnosis of AECOPD in patients presenting to health care facilities, in remote and resource-limited situations, and in circumstances where presentation to healthcare facilities is not possible,” wrote the investigators.
Early detection of AECOPD can decrease the duration of an exacerbation and risk of hospitalization. Delayed detection can lead to a 2-fold increase in risk of hospital admission, largely because patients often wait more than 24 hours after symptom onset before presenting to a hospital. Incorrect AECOPD diagnoses can also result in inappropriate treatments and worsening of symptoms as patients wait longer to receive an accurate diagnosis.
“The early identification and prevention of AECOPD such that patients no longer require hospitalization represent a critical juncture in developing a cost-effective disease management strategy. Rapid identification of AECOPD is imperative to ensure the timely initiation of appropriate and suitable treatment,” said the investigators.
Previous studies have shown that tactics traditionally used to diagnose AECOPD, such as questionnaires and remote spirometry, produce limited efficacy because they significantly rely on patients being able to accurately notice and describe changes in symptoms. Approximately two-thirds of patients have difficulties recognizing worsening of key symptoms associated with AECOPD.
Literature has shown that cough sound recognition technology is effective when used for diagnostic purposes in comparison with other respiratory diagnostic methods. The investigators developed an algorithm aimed at detecting mild AECOPD cases using a smartphone to record 5 coughs and simple patient-reported clinical signs.
For the prospective diagnostic accuracy study, the investigators enrolled 177 patients with COPD. The patients were recruited from the emergency department, low-acuity ambulatory care, and inpatient wards of a large metropolitan hospital in Western Australia.
After excluding inaccessible data, 164 patients were included in the analysis, of whom 78 had COPD and 86 had AECOPD. The patients had a mean (SD) age of 71.8 (10.2) years, 54.9% (n = 90) were female, and 23.2% (n = 38) had heart failure as a comorbidity. Overall, 84% of AECOPD cases were mild.
More patients with AECOPD had heart failure than those with COPD (31.4% vs 14.1%; P = .009). In the AECOPD group, 32 (37%) patients reported experiencing a fever and 60 (70%) reported acute cough.
The algorithm demonstrated high accuracy at diagnosing AECOPD, producing a positive percent agreement (PPA) of 82.6% (95% CI, 72.6-89.9) and a negative percent agreement (NPA) of 91.0% (95% CI, 82.4%-96.3%).
A high diagnostic agreement level was also observed among patients over age 65 years, a subset of the COPD population that is more likely to have comorbidities, demonstrating a PPA of 85.9% (95% CI, 75.0%-93.4%) and an NPA of 88.9 (95% CI, 78.4%-95.4%).
When the investigators plotted the receiver operator curves, the curved produced area under the curve values of 0.89 (95% CI, 0.84-0.94) for all patients and 0.91 (95% CI, 0.86-0.96) for patients over the age of 65.
The investigators said that patients generally found the application easy to use and that it did not exhibit any safety concerns.
The algorithm was accurate at diagnosing AECOPD 79.2% (95% CI, 68.0%-87.8%) of the time in patients with a low risk of 30-day mortality (n = 72). It also correctly diagnosed AECOPD in patients with a moderate risk of 30-day mortality (n = 14).
However, the investigators said that the small number of patients with a moderate risk in the study could skew reporting accuracy. Additionally, the patients included in the analysis were predominately White and had COPD that was a result of smoking history, suggesting that results may not be generalizable to patients with COPD that has a different etiology.
Claxton S, Porter P, Brisbane J, et al. Identifying acute exacerbations of chronic obstructive pulmonary disease using patient-reported symptoms and cough feature analysis. npj Digit Med. July 2, 2021;4(107). doi: 10.1038/s41746-021-00472-x