Could Artificial Intelligence Help With Cancer Diagnoses?

In a year in which one of the most widely-discussed technologies in the medical world has been artificial intelligence, the potential for machine learning and AI’s use to help accelerate analysis and tests by private oncologists is profound.

A few years ago, there was the famous story of how a Japanese AI-powered bread scanner was adapted to help doctors check for cancerous cells, and since then many other attempts have been made to boost the potential of diagnostic equipment through AI.

A suitably trained machine learning algorithm could be used by a doctor to highlight potentially worrying signs for a doctor to check, or provide a second opinion in edge cases if a doctor needs to examine a test further.

This potential has been established in several studies, but the question remains about its use in the wider oncology world.

Ultimately, as with any other medical instrument, AI could be beneficial if not outright revolutionary as long as developers, regulators and oncologists alike are appropriately prepared.

There needs to be a robust training system and continuous professional development to allow skilled medical staff to advance their skills with digital tools and AI, in order to feel confident enough to use them in critical situations.

There also needs to be an interoperable infrastructure such as the Welsh Clinical Portal that allows for imaging databases, electronic patient records and test results to be accessed by AI tools and build up the required context to interpret test results.

It also needs to be a system that makes the lives of patients and doctors easier, which includes clear communications of the capabilities of AI and developing trust that data is only used in ways that patients consent to.

Finally, it needs to narrow the gaps in the healthcare system rather than widen them, which means using diverse datasets to train AI and avoid potential biases, as well as ensuring it is accessible across Sheffield and the United Kingdom.