Session Item

The application of artificial intelligence (AI) in healthcare has attracted extraordinary interest in recent years, and it is commonly acknowledged that AI is going to transform healthcare processes. A number of AI-based applications are currently being researched in several health service fields, and this interest is expected to increase and proliferate in the near future. The "health data ecosystem," which refers to the variety of data that might be used to create AI-based models, is also rapidly growing. Nevertheless, there is widespread concern that digital disparities continue to exist between developed and developing countries and that many low- and middle-income countries (LMICs) lack affordable access to information and communication technologies, particularly in healthcare. Considering the large inequalities between developed and developing countries in terms of access to health care as well as (digital) equipment and proper infrastructure availability, this digital gap might even increase with more AI-based products being introduced into clinical practice. Even though a wide number of processes involving radiation therapy and medical physics have been studied, there are still few AI-based clinical solutions commercially available. The deployment of AI-based tools in healthcare is facing numerous challenges in the technological, safety, ethical, and regulatory domains.

Before considering the introduction of any advanced technological tool into clinical practice, its safety and efficacy should be validated, and a proper risk analysis should be performed. A core team of radiation medicine professionals is required to lead the selection and safe implementation of AI-based tools in the clinic. As a quantitative physical scientist in a clinical setting, the Clinically Qualified Medical Physicist (CQMP) should be considered a key contributor to this process.

The Dosimetry and Medical Radiation Physics (DMRP) Section at the IAEA is working at different levels to provide instruments to CQMPs to be able to safely and effectively deal with AI-based tools in the near future. New guidelines for CQMPs as well as educational and training materials are currently under development with the support of experts in the field.

Regardless of the income level of a country, to ensure a safe application of AI-based technologies in performing a specific medical physics task in a given clinical setting, it is currently recommended that a sufficient number of CQMPs with prior experience with "standard applications" of the task are locally available. Apart from the standard core education and training that is needed to become a CQMP, these physicists should also have adequate theoretical knowledge about the principles of AI as well as the practical skills for selecting, accepting, commissioning, and quality-assuring the AI-based tools.