Skip to main content
ConsultationsConsultations Archive

ISO/PWI 24051-1 Medical laboratories — Part 1: General principles for the application of artificial intelligence in medical laboratories

By November 2, 2023No Comments

Deadline: 28 November

Scope

This document specifies general principles for the application of artificial intelligence in the medical laboratory.

This document is applicable to methods commonly considered subsets of artificial intelligence, including fuzzy logic, Bayesian networks, supervised and unsupervised machine learning, deep learning, neural networks, expert systems, robotics, natural language processing and image analysis.

 

Purpose

Artificial intelligence refers to the use of computers and machines to perform human-like tasks and mimic human cognition in the analysis and interpretation of data. Methods commonly considered subsets of artificial intelligence include fuzzy logic, Bayesian networks, supervised and unsupervised machine learning, deep learning, and neural networks. Expert systems, robotics, natural language processing and image analysis are possible applications of such methods. In this document, artificial intelligence is used as the umbrella term for any or all the subsets. In many instances throughout the document, machine learning could be uses instead of artificial intelligence.

Artificial intelligence methods are increasingly used in laboratory medicine. Applications can range from laboratory utilization to result prediction and validation to result interpretation and clinical decision support. Image analysis is being applied by in vitro diagnostic medical devices and specialties utilizing image data for examination purposes.

Artificial intelligence applications in the medical laboratory can influence the results of laboratory activities and, therefore, need to be considered when managing medical laboratory risk and quality. The continuous use of new data allows artificial intelligence methods, especially machine learning algorithms, to improve their performance over time. Such methods can, and naturally will, change throughout their lifecycles. Currently, there are no guidelines for best practices for determining acceptability criteria, initial validation, evaluating measurement uncertainty, ensuring the validity of results, and risk management of artificial intelligence application in the medical laboratory.

This purpose of this document is to specify general principles for the application of artificial intelligence in the medical laboratory with special consideration of the requirements for quality and competence set forth in ISO 15189:2022. Its creation was recommended at the 27th Meeting of ISO/TC 212, Clinical laboratory testing and in vitro diagnostic test systems (ISO/TC212/Resolution 604). This document is to be accompanied by specific application documents (for example, digital pathology, algorithm-based image analysis). Other interested Technical Committees will be invited to participate.

You can comment via this link.

Ben Kemp