Development of a daily predictive model for the exacerbation of chronic obstructive pulmonary disease

Abstract

Background: Acute exacerbation (AE) of chronic obstructive pulmonary disease (COPD) compromises health status; it increases disease progression and the risk of future exacerbation. We aimed to develop a model to predict COPD exacerbation on a daily basis.

Methods: We merged the Korean COPD subgroup study (KOCOSS) dataset with nationwide medical claims data, information regarding weather, air pollution, and epidemic respiratory virus data. The Korean National Health and Nutrition Examination Survey (KNHANES) dataset was used for validation. Several machine learning methods were employed to increase the predictive power.

Results: The development dataset consisted of 590 COPD patients from the KOCOSS cohort; these were randomly divided into training (n = 581) and internal validation (n = 569) subsets on the basis of the individual claims data. We selected demographic and spirometry data, comorbidities, medications for COPD and hospital visit for AE, air pollution data and meteorological data, and influenza virus data as contributing factors for the final model. Six machine learning and logistic regression tools were used to evaluate the performance of the model. A light gradient boosted machine (LGBM) afforded the best predictive power with an area under the curve (AUC) of 0.935 and an F1 score of 0.653. Similar favorable predictive performance was observed for the 2,151 individuals in the external validation dataset.

Conclusions: Daily prediction of the risk of COPD AE may help patients to rapidly assess their risk of AE and sill guide them to take appropriate intervention in advance. This might lead to reduction of the personal and socioeconomic burdens associated with AE.