An interpretable machine learning model integrating clinical and ct radiomic features for the detection of interstitial lung disease in idiopathic inflammatory myopathies
Authors' conclusion
Our interpretable machine learning nomogram, integrating clinical risk factors and CT radiomics, enables accurate and non-invasive detection of ILD in IIM patients.
Study context
- Population
- Idiopathic inflammatory myopathy (IIM) patients with high-resolution CT (HRCT) scans, with or without interstitial lung disease (ILD)70 with ILD
- Participants
- 115 patients
- Comparison
- Combined clinical-radiomic model vs clinical-only and radiomics-only models
- Outcome
- Detection of ILD presence (discrimination, AUC)
Methodological detailsretrospective cohort, machine learning (LightGBM), five-fold cross-validation, training and independent test sets, nomogram, 1,316 radiomic features from automated lung segmentation of HRCT; feature selection (correlation filtering, mRMR, LASSO) within training set only
- Retrospective cohort
- Machine learning (LightGBM)
- Five-fold cross-validation
- Training and independent test sets
- Nomogram
- 1,316 radiomic features from automated lung segmentation of HRCT; feature selection (correlation filtering, mRMR, LASSO) within training set only
Performance
As reported in the abstract
ILD detection, AUC
| Test or model | AUC (training set, 95% CI 0.792-0.962) | AUC (independent test set, 95% CI 0.791-1.000) |
|---|---|---|
| Combined clinical-radiomic model | 0.877 | 0.898 |
Combined clinical-radiomic model
AUC (training set, 95% CI 0.792-0.962)0.877
AUC (independent test set, 95% CI 0.791-1.000)0.898
Further findings
Calibration and decision curve analysis
- Calibration curves and Decision Curve Analysis (DCA) confirmed high predictive accuracy and clinical net benefit
Independent clinical predictors
- Serum IgG concentration and presence of cough
Key radiomic features (SHAP)
- e.g., wavelet.HLL_glcm_MCC
Original publication
An interpretable machine learning model integrating clinical and ct radiomic features for the detection of interstitial lung disease in idiopathic inflammatory myopathies. Arthritis Res Ther. Published 5 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →