​Development and validation of a Magnetocardiography-Based nomogram for predicting coronary stenosis in patients with normal resting electrocardiograms.

This study aimed to develop and validate the magnetocardiography (MCG)-integrated nomogram to improve CAD prediction in patients with normal ECGs. This prospective cohort study enrolled 421 patients undergoing coronary angiography at the Fourth Affiliated Hospital of Soochow University. Participants with normal resting ECGs were included. Clinical data encompassed demographics, vital signs, laboratory/imaging results, comorbidities, and echocardiographic indices. Magnetocardiography (MCG) recordings were acquired via the Cardiomox MCG-9 system under electromagnetic shielding. Nomogram development integrated LASSO-regularized logistic regression to select predictors. Key MCG parameters and conventional biomarkers were incorporated into the nomogram’s graphical scoring system. Model performance was evaluated through AUC, calibration, and clinical utility (decision curve analysis). Analyses used R v4.2.2 and MSTATA v3.1, with p < 0.05 for significance. The nomogram achieved moderate efficacy, with an area under the curve (AUC) of 0.777 (95% CI: 0.723–0.831) in the training cohort and 0.643 (95% CI: 0.543–0.743) in validation. MCG-derived parameters, including TT_mean_F_30 (OR = 1.01, p = 0.035), identified subclinical repolarization abnormalities undetectable by ECG. Calibration analysis showed moderate accuracy (Brier score = 19.1). External validation revealed the nomogram’s clinical applicability for guiding therapeutic decisions in patients with normal ECGs. This study pioneers the integration of MCG into clinical CAD screening, addressing ECG’s sensitivity limitations. The MCG nomogram offers a scalable tool for early ischemia diagnosis, particularly in resource-limited settings.