Acute coronary occlusion in chest pain patients with broad QRS complex - artificial intelligence-enhanced ECG Interpretation compared to conventional electrocardiographic criteria

Background Identifying acute coronary occlusion (ACO) in chest pain patients with wide QRS complexes remains challenging. Although electrocardiographic criteria exist for left bundle branch block (LBBB) and ventricular pacing (VP), their complexity limits routine use. No standardized ECG-based framework is available for right bundle branch block (RBBB) or non-specific intraventricular conduction delay (NIVCD). Artificial intelligence (AI)–enhanced ECG analysis may provide incremental value, but its performance in these subgroups is unclear. Purpose To evaluate the diagnostic performance of an AI-enhanced ECG model for detecting ACO in chest pain patients with bundle branch block or NIVCD in a metropolitan chest pain unit (CPU). In LBBB and VP, performance was compared with established ECG criteria. Methods The deep convolutional neural network model was previously trained and externally validated in international cohorts. All consecutive chest pain patients presenting between November 2018 and June 2021 with QRS ≥120 ms were included. ECGs were independently classified by two investigators, with adjudication by a third if needed. Clinical management followed European Society of Cardiology guidelines. The endpoint was ACO, adjudicated by an independent and blinded physician using coronary angiography, biomarkers, and echocardiography. Diagnostic performance was assessed by receiver operating characteristic (ROC) analysis. In LBBB and VP, AI model was compared with Sgarbossa, Smith-modified Sgarbossa, and Barcelona criteria. Results Among 439 patients (median QRS 144 ms [132–158]): 192 RBBB (43.7%), 113 LBBB (25.7%), 70 NIVCD (15.9%), and 64 VP (14.6%). ACO prevalence was 3.2% (n = 14: 9 RBBB, 4 LBBB, 1 VP). The AI model detected 8/14 ACO cases (57%) and produced two false positives in the low-risk subgroup (n = 216). Specificity was high in RBBB and NIVCD, with acceptable sensitivity in RBBB. In VP, all cases were correctly classified by AI model, outperforming conventional criteria (PPV 100% vs. 9.1–20%). In LBBB, AI model, Barcelona, and Smith-modified Sgarbossa criteria performed similarly. Conclusion The AI-enhanced ECG model demonstrated high specificity in patients with RBBB, a subgroup in which no established ECG reference standard currently exists, thereby underscoring the need for further evaluation in independent cohorts. In chest pain patients with LBBB, the model achieved diagnostic performance comparable to established yet complex ECG criteria, while showing superior classification accuracy in the setting of ventricular pacing. Subject to external validation in cohorts with a higher prevalence of ACO, this approach may represent a rapid and pragmatically applicable diagnostic alternative.Figure 1Figure 2