ML Multimodal Model with CT Radiomics for Interstitial Lung Disease Prediction: External Validation

Multimodal ML model integrating CT radiomics and clinical features predicted ILD with AUC 0.715 (95% CI 0.661–0.770) on external validation (614 patients) and 0.733 on internal validation.

• Multicenter retrospective study with 2456 subjects; development cohort n=1842, external validation cohort n=614.
• Multimodal model outperformed single-modality radiomics or clinical models.
• More complex ML algorithms did not improve external validation performance.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.