In Chest ; h5-index 81.0
RATIONALE : Asthma exacerbations result in significant health and economic burden but is difficult to predict.
OBJECTIVES : We aim to develop machine learning (ML) models with large-scale outpatient data to predict asthma exacerbations.
METHODS : We analyzed data extracted from electronic health records (EHR) on asthmatics followed at the Cleveland Clinic from 2010 to 2018. Demographic information, comorbidities, lab values and asthma medications were included as covariates. Three different models were built with Logistic regression, random forest, and gradient boosting decision tree to predict: (1) non-severe asthma exacerbation requiring oral glucocorticoid burst, (2) emergency department (ED) visits and (3) hospitalizations.
MEASUREMENTS AND MAIN RESULTS : Out of 60,302 patients, 19,772 (32.8%) had at least one non-severe exacerbation requiring oral glucocorticoid burst, 1,748 (2.9%) requiring ED visit, and 902 (1.5%) requiring hospitalization. Non-severe exacerbation, ED visit and hospitalization were best predicted by Light Gradient Boosting Machine (LightGBM), an algorithm used in machine learning to fit predictive analytic models, and had an area under the receiver operator curve of 0.71 (95%CI: 0.70-0.72), 0.88 (95%CI 0.86-0.89) and 0.85 (95%CI: 0.82-0.88) respectively. Risk factors for all three outcomes include age, long acting beta agonist, high dose inhaled glucocorticoid or chronic oral glucocorticoid therapy. In subgroup analysis of 9,448 patients with spirometry data, low FEV1 and FEV/FVC ratio were identified as top risk factors for asthma exacerbation, ED visits and hospitalization. However, adding pulmonary function tests did not improve models' prediction performance.
CONCLUSIONS : Models built with ML algorithm from real-world outpatient EHR data accurately predict asthma exacerbation and can be incorporated into clinical decision tools to enhance outpatient care and prevent adverse outcomes.
Zein Joe G, Wu Chao-Ping, Attaway Amy H, Zhang Peng, Nazha Aziz