Explainable Biological Age from Automated Chest Radiography-based Organ Quantifications: qCXR-bioage

Area of Science:

• Radiology
• Artificial Intelligence
• Biomarkers

Background:

• Assessing biologic age is crucial for predicting health outcomes.
• Quantitative organ metrics from chest radiographs offer potential biomarkers.
• Existing methods for biologic age estimation have limitations.

Purpose of the Study:

• To develop an explainable biologic age model (qCXR-bioage) using quantitative chest radiograph metrics.
• To evaluate the prognostic value of qCXR-bioage for all-cause and cause-specific mortality.
• To validate the model's performance in a large, independent cohort.

Main Methods:

• Developed qCXR-bioage using deep learning to extract biomarkers (lung area, emphysema, aortic diameter, heart area, bone density) from chest radiographs.
• Integrated automated measurements with clinical variables using the Klemera-Doubal method and multivariable ridge regression.
• Validated the model in 257,004 individuals and assessed mortality prediction using Cox and Fine-Gray models.

Main Results:

• qCXR-bioage strongly correlated with chronological age (R² > 0.98).
• The model outperformed chronological age in predicting all-cause mortality (C-index > 0.76).
• Accelerated aging (qCXR-bioage > chronological age) was significantly associated with increased all-cause and cardiovascular mortality.

Conclusions:

• Chest radiograph-derived qCXR-bioage is a reliable and explainable tool for estimating biologic age.
• qCXR-bioage demonstrates significant prognostic value for predicting mortality, including cardiovascular causes.
• This approach offers a novel method for risk stratification using conventional radiography.