Development and Evaluation of an Algorithm to Detect COPD Through an Intelligent Device

Abstract Rationale:Chronic obstructive pulmonary disease (COPD) is one of the most common respiratory diseases. Early detection and treatment are critical to prevent the deterioration of COPD. In this study, we aim to establishment and evaluation an algorithm that can detect COPD from physiological parameters and audio data which are collected by a wearable device.Methods: Participants were instructed to use HUAWEI smart wearable devices to collect physiological data, including oxygen saturation, heart rate, respiratory rate, and cough sounds. The collected data served as measurement inputs. To evaluate the accuracy of the HUAWEI smart wearable device and its associated algorithms for lung function assessment and COPD screening, a comparative analysis was conducted. This analysis used the results of medical spirometers and the diagnoses of hospital respiratory specialists as the gold standard for assessing lung function and diagnosing COPD.Results:400 subjects (236 males,164 females; 200 with stable COPD and 200 without COPD) were recruited for algorithm development, among COPD patients, age distribution was≤50 years(30), 51-65years(73), and > 65years(97), with GOLD stages 1(58), 2(102), and 3+4(40). A further 400 subjects (240 males, 160 females;200 with COPD and 200 without COPD) were recruited for testing the algorithm.The overall mean age was 57.8 years, mean BMI was 24.7kg/m2, and mean FEV1/FVC ratio was 0.69.Pulmonary Function results showed an MAE of 6.6% for FEV1/FVC and 10.0% for FVC.COPD screening achieved 80.8% accuracy,80.0% sensitivity and 81.5% specificity.Conclusions:The test results based on Huawei's smart wearable device and its accompanying lung function assessment and COPD screening algorithms are highly consistent with the diagnostic results of medical spirometers and professional doctors, and are effective for lung function assessment and COPD screening.