Application of Artificial Intelligence as a Support Tool in Internal Audits of Clinical Laboratories Accordingto ISO 15189:2022

Abstract

Introduction: Internal audits in clinical laboratories are a fundamental quality component according to ISO 15189:2022 “Medical laboratories — Requirements for quality and competence”. Artificial Intelligence (AI) shows potential as a support tool for this process. Objective: To evaluate the use of Large Language Models (LLMs) as a tool for technical writing, classification (nonconformity/compliance/opportunity for improvement), and normative referencing of internal audit findings based on ISO 15189:2022, within a structured methodological framework validated by expert auditors. Materials and Methods: An LLM (ChatGPT) was employed, using a structured interaction strategy based on the ASPECCT method. Tool performance was assessed using 39 previous internal audit findings. Two human auditors independently evaluated the AI using predefined criteria: finding classification, technical writing, normative referencing, compliance with constraints, and capacity for improvement through successive iterations. Results: The AI interaction was successfully developed within the ASPECCT methodological framework. Expert auditors demonstrated high inter-rater agreement: 97.65% for the overall assessment and 100% for finding classification. In addition, 99.23% of the ratings were categorized as “good” (21.37%) or “excellent” (77.86%), and 64.1% of the initial AI results required no corrections. Discussion: AI, within a structured methodological framework and under human supervision, constitutes a valuable tool for internal audits, contributing to increased standardization and time optimization.