AUTHOR=Wang Jennifer M. , Labaki Wassim W. , Murray Susan , Martinez Fernando J. , Curtis Jeffrey L. , Hoffman Eric A. , Ram Sundaresh , Bell Alexander J. , Galban Craig J. , Han MeiLan K. , Hatt Charles 

TITLE=Machine learning for screening of at-risk, mild and moderate COPD patients at risk of FEV1 decline: results from COPDGene and SPIROMICS

JOURNAL=Frontiers in Physiology

VOLUME=14

YEAR=2023

URL=https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2023.1144192

DOI=10.3389/fphys.2023.1144192

ISSN=1664-042X

ABSTRACT=<p><bold>Purpose:</bold> The purpose of this study was to train and validate machine learning models for predicting rapid decline of forced expiratory volume in 1 s (FEV<sub>1</sub>) in individuals with a smoking history at-risk-for chronic obstructive pulmonary disease (COPD), Global Initiative for Chronic Obstructive Lung Disease (GOLD 0), or with mild-to-moderate (GOLD 1–2) COPD. We trained multiple models to predict rapid FEV<sub>1</sub> decline using demographic, clinical and radiologic biomarker data. Training and internal validation data were obtained from the COPDGene study and prediction models were validated against the SPIROMICS cohort.</p><p><bold>Methods:</bold> We used GOLD 0–2 participants (<italic>n</italic> = 3,821) from COPDGene (60.0 ± 8.8 years, 49.9% male) for variable selection and model training. Accelerated lung function decline was defined as a mean drop in FEV<sub>1</sub>% predicted of &gt; 1.5%/year at 5-year follow-up. We built logistic regression models predicting accelerated decline based on 22 chest CT imaging biomarker, pulmonary function, symptom, and demographic features. Models were validated using <italic>n</italic> = 885 SPIROMICS subjects (63.6 ± 8.6 years, 47.8% male).</p><p><bold>Results:</bold> The most important variables for predicting FEV<sub>1</sub> decline in GOLD 0 participants were bronchodilator responsiveness (BDR), post bronchodilator FEV<sub>1</sub>% predicted (FEV<sub>1</sub>.pp.post), and CT-derived expiratory lung volume; among GOLD 1 and 2 subjects, they were BDR, age, and PRM<sub>lower lobes fSAD</sub>. In the validation cohort, GOLD 0 and GOLD 1–2 full variable models had significant predictive performance with AUCs of 0.620 ± 0.081 (<italic>p</italic> = 0.041) and 0.640 ± 0.059 (<italic>p</italic> &lt; 0.001). Subjects with higher model-derived risk scores had significantly greater odds of FEV<sub>1</sub> decline than those with lower scores.</p><p><bold>Conclusion:</bold> Predicting FEV<sub>1</sub> decline in at-risk patients remains challenging but a combination of clinical, physiologic and imaging variables provided the best performance across two COPD cohorts.</p>