Decision Tree-Based Model Prediction of Lower Extremity CTA Angiography Using Laboratory Biomarkers in Young Takayasu Arteritis Patients

Takayasu arteritis (TAK) is a rare, chronic large vessel vasculitis that primarily affects young patients, leading to arterial wall thickening, stenosis, or occlusion with potentially severe complications. Computed tomography angiography (CTA) is a key diagnostic tool but involves risks due to contrast media and radiation, particularly in young populations. This study aimed to develop and evaluate a decision tree model using routine laboratory biomarkers to predict lower extremity arterial abnormalities detected by CTA in young patients suspected of TAK. We retrospectively analyzed 194 patients under 40 years old with lower limb CTA and comprehensive laboratory data, excluding confounding comorbidities. Ten laboratory parameters including hs-CRP, LDH, HDL, creatine kinase, coagulation factors, IL-6, ESR, and albumin were incorporated into a decision tree classifier. The model achieved 80% accuracy, sensitivity, and specificity in internal cross-validation. External validation on 20 patients showed 80% accuracy, outperforming an expert radiologist's 55% accuracy based on laboratory data alone (p < 0.05). Albumin was the only individual biomarker significantly different between groups, highlighting the utility of integrating multiple biomarkers rather than relying on single parameters. The decision tree model offers a transparent, noninvasive, and safer adjunct or alternative to CTA for early detection of arterial involvement in young TAK patients, potentially reducing unnecessary imaging and associated risks. Further prospective multicenter validation is warranted to confirm its clinical applicability and integration into diagnostic workflows.