Repolarization in bundle branch block, can artificial intelligence offer insights?

Abstract Background QT interval interpretation remains challenging in patients with bundle branch block (BBB), where QRS widening prolongs the QT interval due to delayed ventricular depolarization. A validated correction formula, QTm = QTBBB − (50% × QRSBBB), compensates for prolongation of depolarization time due to BBB, , enabling estimation of true repolarization time. However, its clinical impact on automated ECG prioritization systems requires validation. Objective To evaluate the effect of BBB-specific QT correction on automated severity classification and validate algorithm performance in detecting BBB and measuring intervals. Methods We analysed 250 ECGs selected from a proprietary database: 50 BBB cases (25 LBBB, 25 RBBB) and 200 controls (125 with normal conduction, 75 with non-specific QRS widening ≥110 ms). QTBBB and QRSBBB were defined as the longest globally measured QT and QRS intervals on the 12-lead ECG. The PulseAI neural network classified BBB and measured intervals. QT correction (QTm) was applied to BBB cases, followed by Fridericia’s rate adjustment. Severity classifications included: Emergent (≥500 ms), Significant (≥470 ms), Borderline (≥430 ms), and Normal (