小児外科における人工知能統合の倫理的および法的課題

Artificial intelligence (AI) technologies are increasingly being trialled with applications spanning imaging, robotic assistance and early risk prediction. Children's unique legal status presents distinct ethical and legal concerns. The objective of this review was to outline key ethical and legal challenges from AI integration into paediatric surgery and propose practical strategies for clinicians and policymakers.

This article is a narrative review of peer-reviewed and grey literature (2015-2025), and statutory guidance (including the Information Commissioner's Office [ICO] Children's Code, Medicines and Healthcare products Regulatory Agency [MHRA] Software as a Medical Device Roadmap, General Medical Council guidance), and documentation from paediatric AI hubs. Ethical analysis was framed using principlism (autonomy, beneficence, non-maleficence, justice), with modifications for paediatric contexts (evolving capacity, best interests).

Five core challenges were identified: safeguarding sensitive paediatric data; facilitating informed consent; mitigating algorithmic bias; clarifying liability in adaptive systems; and harmonising regulatory oversight. Recommendations include federated learning networks, transparent model cards, a pooled indemnity fund, AI literacy training and ongoing performance audits.

AI holds promise for safer, more equitable paediatric surgery with robust ethical safeguards, updated legal frameworks and sustained regulatory vigilance. What is already known: (i) AI enhances image guidance, robotic precision and risk prediction in surgery; (ii) children's data require special protection under the UK General Data Protection Regulation and ICO Children's Code; and (iii) NHS England funds AI pilots, but paediatric-specific governance remains underdeveloped. This study: (i) maps UK-specific ethical (privacy, consent, bias) and legal (Montgomery, Consumer Protection Act 1987, MHRA) duties in AI-assisted paediatric surgery; (ii) proposes a federated data-sharing consortium and tiered liability model, and offers a practical governance framework for clinicians, regulators, and NHS boards.