UK Develops a New Framework for AI Systems in Health Care

The British Standards Institution (BSI) – the U.K.’s national standards body − published a new framework to foster greater trust in AI systems and products for health care.

The auditable standard, “Validation framework for the use of AI within healthcare – Specification (BS 30440),” can serve as a guide to developers of AI systems used to diagnose or treat patients, or manage their health.

AI tools such as at-home monitoring gadgets, medical devices and smartphone chatbots can help clinicians and other health providers make informed diagnostic decisions more efficiently. But providers do not always have the skills, budget and time to assess AI-powered medical products.

The new standard is meant to show that the AI systems were developed in an ethical safe, and effective way. It also helps providers evaluate AI products in terms of integration into the clinical environment, standards of performance, clinical benefits, socially equitable outcomes and ethical behavior.

The new standard fills a regulatory gap: There currently is no legislation governing the use of AI in health care.

BSI said the framework can be audited to check AI systems for conformity, which can lead to certification. Health care organizations can then require that AI systems have this certification.

The standard applies to the following:

The specifications provide guidance for systems, models, products or technologies that use machine learning and other AI techniques.

The standard was created by a group of experts, including AI specialists, software engineers, clinicians, health care leaders, and ethicists. They identified best practices based on existing guidance standards, which were then boiled down to an assessment of complex functionality.

BSI said it has been working with regulators, health care groups and others to look at the role of standards, as regulations and governance develops around AI in health care.