BACKGROUND
World Health Organization anticipated that by 2030, chronic obstructive pulmonary disease (COPD) will be the third leading cause of mortality and the seventh leading cause of morbidity worldwide. Acute exacerbations of chronic obstructive pulmonary disease (AE-COPD) are associated with accelerated decline in lung function, diminished quality of life, and higher mortality. Accurate early detection of acute exacerbations will enable early management and reduce mortality.
OBJECTIVE
To develop a prediction model of AE-COPD using lifestyle data, environment factors and patient’s symptoms to achieve early detection of AE-COPD in the forthcoming 7 days.
METHODS
This prospective study was conducted in National Taiwan University Hospital. COPD patients without pacemaker and pregnancy were invited for enrollment. Lifestyle, temperature, humidity and fine particulate matter (PM2.5) were collected using wearable devices, home air quality sensing devices, and smartphone application. The episodes of AE-COPD were evaluated by standardized questionnaires. With these input features, we evaluated the prediction performance of machine learning models with random forest, decision tree, kNN, linear discriminant analysis, AdaBoost, and a deep neural network model.
RESULTS
The continuous real-time monitoring of lifestyle and indoor environment factors were implemented in this study by integrating home air quality sensing devices, smartphone applications, and wearable devices. All data from 67 COPD patients were collected prospectively during a mean of 4-month follow-up and 25 episodes of AE-COPD were detected. For prediction of AE-COPD within the next 7 days, our AE-COPD predictive model had accuracy of 92.1%, sensitivity of 94%, and specificity of 90.4%. The receiver operating characteristic curve analysis showed the area under the curve of the model in predicting AE-COPD was >0.9. The most weighting variables in the model were daily walking steps, climbing stairs and daily moving distances.
CONCLUSIONS
Using wearable devices, home air quality sensing devices, smartphone application and supervised prediction algorithms, we achieved an excellent predictive power for the task of predicting whether a patient will experience an acute exacerbation of COPD within the next 7 days. The system was capable of making reliable predictions with enough time in advance when a patient is going to have an AE-COPD.
Machine learning algorithms and forced oscillation measurements applied to the automatic identification of chronic obstructive pulmonary disease
