Deep Learning-Based Estimated Pulmonary Biological Age From Chest Computed Tomography Images in Healthy Adults: Model Development and Validation Study.

Background: Estimated pulmonary biological age (ePBA) has emerged as a more reliable indicator for disease progression and mortality than chronological age, with chest computed tomography (CT) as a promising tool for calculating ePBA. However, the lack of models trained and validated with large-scale healthy adults hinders the generalizability of the CT-based ePBA. Objective: This study aims to develop an aging biomarker (ePBA) from multicenter chest CTs of healthy adults using deep learning and investigate the association between age gap (ePBA - chronological age) and pulmonary function as well as all-cause mortality in patients with chronic obstructive pulmonary disease (COPD). Methods: We used 11,187 chest CT scans from healthy adults at 3 health management centers and used multiple deep learning models. Of these, 7726 scans from institution A were used for model development. The remaining CT scans from institutions B (n=1506) and C (n=1955) served as external test datasets. To examine whether ePBA provided information beyond chronological age in patients with the disease, we investigated the association of age gap with lung function and all-cause mortality among 138 patients with COPD hospitalized at the same time period in institution A. Results: The deep learning models demonstrated acceptable applicability for this task and exhibited a strong correlation between ePBA and chronological age. Age gap was significantly associated with forced expiratory volume in 1 second expressed as percentage of predicted values reduction (rs=-0.18; P=.03) and an increased risk of all-cause mortality (hazard ratio: 1.16, 95% CI 1.08-1.25) in patients with COPD. Conclusions: This study developed and validated a biomarker of aging-ePBA-with deep learning models based on chest CT. Age gap could serve as a novel clinical biomarker in patients with COPD.