Artificial intelligence for testing and calibration laboratory systems: a structured systematic review and ISO/IEC 17025-oriented governance framework

High Energy Chemistry · Springer Nature · 2026-08-29 · DOI: 10.1007/s00769-026-01735-4

Аннотация

Artificial intelligence (AI) is increasingly being explored for laboratory testing, calibration, data management, and decision support. For laboratories operating within ISO/IEC 17025-oriented quality systems, however, AI adoption raises specific questions about validation, metrological traceability, measurement uncertainty, data integrity, impartiality, and human accountability. This study is presented as a structured systematic review with thematic synthesis and ISO/IEC 17025 clause mapping. Thirty peer-reviewed studies were included in the qualitative evidence synthesis, while metrological, accreditation, and AI governance documents were used as contextual sources for standards-oriented interpretation. The review shows that most available evidence comes from laboratory-adjacent settings such as clinical diagnostics, environmental sensor calibration, laboratory information systems, and AI governance, rather than from laboratories explicitly accredited to ISO/IEC 17025. The strongest technical evidence relates to analytical pattern recognition and sensor calibration, whereas direct evidence for accreditation governance, uncertainty treatment, and routine operational validation remains limited. The principal gap is the lack of standardized validation and governance frameworks for deploying AI in testing and calibration environments. The proposed framework therefore positions AI as a controlled support tool rather than an autonomous decision-maker and links each AI use case to documented validation, GUM-based uncertainty evaluation where applicable, traceability evidence, change control, audit trails, and authorized signatory review. The paper contributes a clause-by-clause mapping and an operational lifecycle for AI adoption in ISO/IEC 17025-oriented laboratory systems.

Читать у издателя →

Похожие статьи