Ethical integration of artificial intelligence in pediatric epilepsy neuropsychology: decision authority,...

Artificial intelligence (AI) is increasingly being incorporated into pediatric epilepsy care, with applications in surgical outcome prediction, cognitive risk estimation, and prognostic modeling. Although these technologies offer opportunities to improve clinical decision support through integration of multimodal data, their ethical implementation within pediatric epilepsy neuropsychology remains insufficiently defined. This narrative review, conducted in accordance with the Scale for the Assessment of Narrative Review Articles (SANRA), synthesizes literature from neuropsychology, epilepsy, pediatric ethics, and explainable AI to examine the role of AI in pediatric epilepsy surgical decision-making. Three interrelated domains are explored: (1) decision authority, (2) communication with patients and families, and (3) ethical considerations in surgical risk modeling. I argue that the central challenge is not whether AI can generate accurate predictions, but how those predictions are interpreted, communicated, and incorporated into decisions involving uncertainty, developmental vulnerability, and long-term outcomes. To address this challenge, I propose three conceptual frameworks: a three-tier model of decision authority, a framework distinguishing statistical and ethical thresholds for action, and a workflow model for AI-augmented clinical communication. Across these domains, neuropsychologists serve as interpreters of developmental meaning, communicators of uncertainty, and facilitators of value-sensitive decision-making. Ethical integration of AI therefore requires preserving clinician accountability while using predictive technologies as augmentative rather than autonomous tools. These proposed frameworks are intended to guide future discussion, clinical implementation, and empirical evaluation rather than represent established or validated models of practice.