Development of a Screening Model for Exercise-Induced Desaturation by Machine Learning Method.

backgroundExercise-induced desaturation (EID) during the 6-minute walk test (6MWT) is an established marker of adverse outcomes in patients with chronic obstructive pulmonary disease (COPD). Therefore, we aimed to develop a machine learning approach focused on screening to identify patients at increased risk of EID.

methodsWe analyzed data from the nationwide, multicenter Korea COPD Subgroup Study. EID was defined as a peripheral oxygen saturation (SpO2) <90% with a decrease of ≥4%p. The cohort was divided into training (80%) and test (20%) sets. Candidate predictors were selected using the Boruta algorithm, and models were developed using multivariable logistic regression, extreme gradient boosting (XGB), random forest, and support vector classification, employing a screening-oriented threshold strategy that prioritized sensitivity.

resultsAmong 1,788 patients with COPD, 185 (10.3%) exhibited EID. All models demonstrated a high precision-recall area under the curve (PR-AUC) during internal validation. The predictors selected by the Boruta algorithm included body mass index, COPD Assessment Test, St. George's Respiratory Questionnaire for COPD patients, mental health indicators, pulmonary function parameters, X-ray-identified bronchiectasis, and hemoglobin level. In the independent test set, PR-AUC decreased across models, while calibration metrics showed modest differences between datasets. The XGB model achieved the highest sensitivity during internal validation, and both its sensitivity and specificity remained relatively stable in the test set. Baseline SpO2 and diffusion capacity of the lung for carbon monoxide were the most influential predictors.

conclusionA screening-oriented machine learning approach utilizing routinely available variables may facilitate targeted referral for the 6MWT in COPD patients.