Research highlights the need for sustainable AI solutions in healthcare

As in other sectors around the world, the rapid development of artificial intelligence (AI) has found its place in healthcare, especially in the field of radiology. As such, AI-based diagnostic systems are flourishing, and hospitals are rapidly adopting the technology to assist radiologists. On the other hand, there are concerns about the environmental impact of increasingly complex AI models and the need for more sustainable AI solutions.

That’s why Associate Professor Daiju Ueda of the Graduate School of Medicine at Osaka Metropolitan University, a member of the Japan Radiological Society, led a research team in a study of the environmental costs of AI. In this review of research, leading members of the Japan Radiological Society and medical researchers discussed the energy consumption of AI systems in the medical field, carbon dioxide emissions from data centers, and electronic waste issues. Specific solutions to mitigate these environmental effects were discussed, including developing energy-efficient AI models, implementing green computing, and using renewable energy.

In addition, the review proposes measures for the sustainable implementation of AI in the medical field. These are important guidelines for medical institutions, policymakers and AI developers to operate AI systems in an environmentally responsible manner.

AI has the potential to improve the quality of healthcare, but at the same time its impact on the environment cannot be ignored. The best practices we recommend are the first steps towards balancing these two factors. The challenge for the future will be to verify and further develop these recommendations in real medical practice. It is also expected that they will contribute to the standardization of methods for assessing the impact of AI on the environment and the development of an international regulatory framework.”

Daiju Ueda, Associate Professor, Faculty of Medicine, Osaka Metropolitan University

The results were published in Interventional diagnostics and imaging.