Abstract
BACKGROUND: Artificial intelligence (AI)-guided spatiotemporal dispersion (stD) mapping has been shown to improve outcomes in patients with persistent atrial fibrillation (AF). However, the relationship between stD mapping and markers of atrial cardiomyopathy, dispersion patterns in paroxysmal versus persistent AF, stability of dispersion regions, and stD-guided ablation-related outcomes in all-comer cohorts remain elusive. METHODS: In this retrospective single-center analysis, AF patients underwent high-density electroanatomical mapping alongside multiple instances of stD mapping using VOLTA AF Explorer software. Pulmonary vein isolation (PVI) and targeted ablation of left atrial dispersion regions were performed. Clinical, echocardiographic, biomarker, and low-voltage area (LVA) data were collected as markers of left atrial remodeling. RESULTS: stD mapping identified dispersion in 92% of patients. Mean time since AF diagnosis was 7 ± 1 years. Overall, 58% of patients showed dispersion exclusively co-localizing with low-voltage areas, while 42% had dispersion extending into intermediate or normal voltage regions. Dispersion burden correlated strongly with LVA extent and other remodeling markers such as NT-proBNP and LAVI. Persistent AF patients exhibited a significantly higher number of dispersion sites compared to paroxysmal AF. Dispersion patterns remained largely consistent before and after cardioversion in persistent AF, with the posterior left atrial wall emerging as a common hotspot. At follow-up, AF recurred in 33% of paroxysmal and 60% of persistent AF patients who had dispersion ablation limited to the left atrium. Despite these recurrences, most patients reported an improvement in symptomatic burden. CONCLUSION: AI-guided stD mapping effectively identifies atrial remodeling beyond classical voltage-derived substrate, supporting its potential as a useful adjunctive tool in AF characterization.