cAPM: Continual AI-Assisted Pace-Mapping with Active Learning

Short summary

cAPM introduces a continual learning system that transfers knowledge across multiple cardiac ablation targets, using active learning to guide pace-mapping. Achieves 81% clinical accuracy using 4.5 pacing sites versus 38% baseline requiring 13.7 sites. Demonstrates significant efficiency gains through task-agnostic surrogate modeling and sequential knowledge transfer.

• •Novel continual learning approach transfers knowledge across multiple VT targets within and across patients
• •81% localization accuracy vs 38% baseline—4.5 vs 13.7 pacing sites needed
• •Active learning + surrogate modeling reduces clinical procedure time while improving targeting precision

Generated with AI, which can make mistakes.