Helping guide AI ethics in oncology and beyond

How industrial AI can support AI in oncology

For Dr. Matthew Katz, MD, a typical working day can involve helping those undergoing cancer care make life-changing – and sometimes lifesaving – decisions. His patients place an enormous amount of trust in him and his team to do what’s in their best interests and, similarly, Dr. Katz relies on the technology that supports him to do this work.

When he heard a BBC radio piece on the use of ethics in artificial intelligence (AI) on his drive to work one mid-December morning in 2020, his interests were piqued. Just like many applications of AI, its trustworthiness has always been a challenge within medicine, with many examples of systems being shown to reproduce existing biases. Despite the AI’s best intentions, it could be making critical decisions about a person’s care based on data sets that incorporate bias.

The Alethia Framework

What Dr. Katz was learning, however, was about a new framework developed by our data teams at Rolls-Royce that went beyond existing theoretical guidance to focus on the practical application of responsible and trustworthy AI: going from the ‘what?’ to the ‘how?’. He reached out to us at once to discuss collaboration.

What struck Dr. Katz about The Aletheia Framework was the ability to have a transparent view of the data being used to inform an AI’s decision making. This transparency meant that Dr. Katz and other doctors could be informed about whether these decisions were being made with this patient’s best interests at their core and involved the clinical judgement of the medical professional where other frameworks fell short. It seemed that there was a great deal these two industries could learn from one another.

Thoughts to be shared

While both Rolls-Royce’s power technologists and oncology professionals are experts in our respective domains, we also share the characteristic that the public is frequently our end-user and we both need the public to trust our decision making, products and services to keep them safe.

Anyone who’s travelled long-haul in the last 20 years will have, perhaps unknowingly, benefited from the AI and data analytics used to ensure our Trent engines on the aircraft wings are running at their optimum. Not to mention the gigantic data sets created through real-time engine health monitoring to accurately minimise disruptions, maximise availability and secure revenue. These monitoring and quality assurance measures would be invaluable in critical software systems for treatment planning and radiotherapy department workflow.

Additionally, the framework considers factors like social impact. For example, the loss of skills as AI takes on more of the tasks performed by doctors, dosimetrists, physicists and radiographers may mean that additional, new training is needed to help professionals work alongside AI in different ways. As the AI improves to 99.9% accuracy, monitoring the 0.01% will be a key factor in these roles.

Putting AI to the test

Dr. Katz and a team of other oncology professionals joined us in investigating how to maximise the therapeutic ratio between the benefits and harms of AI. As well as co-authoring a paper on safe and ethical AI in radiotherapy, our continued learnings from working together with oncologists – and other industries like education and even music – enabled us to publish a second, more universal iteration of The Aletheia Framework which was published in December 2021.

As AI continues to transform varying sectors, collaboration across industries to share ideas and learn – even unlikely partnerships between engineering and oncology – could be the key to optimising this transition. Within healthcare alone, we’re already pioneering new AI technologies using The Aletheia Framework which could, one day, go as far as using facial recognition to detect signs of post-partum depression. And we’ve already put our expertise to use helping people living with Motor Neurone disease to be able to maintain their voice digitally.

The opportunity to learn from each other is a great one, and we look forward to ongoing collaborations.

US and UK AI oncology experts worked with us to adapt and create their own proposal for an ethical AI framework. Watch the full story

Register for news and stories

Our Alert service delivers the latest press releases, stories and regulatory news directly to your mailbox.