RCR position statement on artificial intelligence

Friday 20 July 2018

Introduction

Artificial Intelligence (AI) has greatly increased in prominence in the last decade, and its use in radiology and clinical oncology within healthcare has attracted an enormous amount of interest from many sources, ranging from government to industry.

The RCR believes that AI potentially represents one of the most fundamental changes in medical care since the inception of the NHS, and strongly welcomes the introduction of appropriately regulated and governed uses of AI related technologies to enhance clinical practice.

Key Principles

The RCR has a vital role to play in the development and integration of AI across medical imaging pathways. For the RCR specialities of clinical radiology and clinical oncology, this role is best considered in context of the following areas:

Patient pathways

AI and related technological advances offer huge opportunities in terms of patients being enabled to self-manage, while communicating and interfacing remotely with their multi professional clinical team. This enables radiologists and oncologists to increase and improve the quality of their time available to provide clinical care.  It also offers potential for vital information such as test results to be communicated in a timely and effective way - empowering and promoting patient self-management.

Additionally, patient reported outcomes can be collected and analysed effectively from real time patient experience, improving clinical knowledge and supporting research, which gives opportunity to shape clinical service requirements for the future.  Patient information can be automatically triangulated and communicated through different media, ensuring immediate accessibility to all service users.  With this technology comes responsibility to ensure that regulatory requirements are met so that patients and their data are protected.

Data

The development of Machine Learning (ML) or Deep Learning (DL) algorithms to enhance radiological and clinical oncology practices requires huge volumes of patient data. These data need to be anonymised, cleansed, labelled and validated. Patients and the general public need to be assured that their data will be handled securely and sensitively.

Regulation

A robust regulatory framework for the integration of AI into medical practice needs to be drawn up.  Many different companies of varying sizes are developing AI tools for use in radiology and clinical oncology. These companies are making claims about the power of these tools - some of which are unsubstantiated.  If tools fail to live up to these claims, public trust in the technology could be damaged.  The RCR believes that regulation and standard setting is vital to ensure that the medical profession and the public alike continue to have confidence in the use of AI in clinical practice across RCR specialties.

Quality Assurance/Governance/Veracity

Introducing AI is an evolutionary process rather than a single revolutionary event; all elements of AI will need to be tested to ensure accuracy, reliability and Quality Assurance of the whole field.  The use of the technology will need appropriate governance.  Published results for sensitivity and specificity of AI tools will be necessary prior to the introduction of any technology in the radiology/clinical workflow.

Training and development

Appropriate training needs to be provided for those radiologists and oncologists who will use AI in their clinical practice. Educational tools will need to be developed to ensure the use and oversight of the technology by clinicians occurs in a safe and robust environment.

Conclusion

Far from making the clinical radiologist and clinical oncologist of the future redundant, the use of AI will help standardise many aspects of clinical care, will optimise processes, and allow greater use of clinical data to inform best practice and outcomes.

Clinical radiologists will work with AI integrated into their daily workflow to improve efficiency and also to increase the scope of their diagnostic capacity, releasing time for direct patient care and for vital research. Clinical oncologists will be enabled to improve patient outcomes through targeted and intelligent administration of radiotherapy, genomics mapping, and biotechnology advances.

As we expand our use and knowledge of the potential of AI in clinical practice, the RCR recognises that these advances force a need to be aware of the capacities and limits of such technologies, as well as the need to be able to critically evaluate the benefits.  This will enable continued delivery of high quality patient care.

The RCR will work with leaders and experts in the field to support the integration of adequately regulated and governed AI as we grow a new generation of radiologists and oncologists.