It's not a matter of if artificial intelligence (AI) is going to become an everyday reality, but when. The question facing the medical profession is how much it can be utilised and what risks are involved.
The mere mention of AI can evoke dystopian visions of machine revolutions and robot uprisings, and when applied to healthcare – where it may literally be involved in life-and-death decisions – AI can appear especially scary and present a number of challenges. The reality though is slightly less terrifying as it's more likely to come in the form of a mobile app that will help to diagnose heart disease, and interpret results and scans.
But it's not a matter of if this will happen, it's happening right now.
This is why the RCP this week published a statement, along with recommendations, on how AI should be used to support doctors deliver healthcare safely. And on Tuesday, a new code of conduct for AI was announced by the Department of Health and Social Care (DHSC) – which set out 10 principles for how AI technology and suppliers interact with NHS patient data.
We welcome this code, which echoes our statement in calling for AI to be used safely and responsibly, and to be viewed as an opportunity for progress and not a threat to existing ways of working
The question facing the medical profession is how much [AI] can be utilised in life-or-death situations, and what risks are involved.
What is AI?
It can be hard to nail down a singular definition of AI as it refers to more than one concept, and plays a role in more areas than just health. In its simplest terms, it’s the simulation of human intelligence by machines. Typically though, it analyses large sets of data and uses that to perform a function via an algorithm or model.
The potential for AI can appear massive with media focus often on many big exciting futuristic projects. However, it’s important to remember AI can also help to solve smaller, real-world challenges – reducing prescribing errors or ensuring patients take their medication, for example.
One of the common concerns is that AI will replace doctors, but this seems unlikely in our lifetimes (particularly in view of the IT problems at a very basic level we experience on a daily basis in the NHS). However, AI is already, and will continue to be, a useful tool to support the work of doctors and help provide the best quality care for patients.
Technology can support doctors to focus on strengthening the relationship between patient and doctor, support shared decision making based on evidence, and take a more holistic view of patient care.
But it’s important we get the guidelines right so that it is used correctly and safely, for example over the use of data.
AI is already, and will continue to be, a useful tool to support the work of doctors and help provide the best quality care for patients.
What do we want from AI?
Our statement, released this week, tackles all sorts of opportunities and potential issues around AI, to try and ensure we in healthcare are ready to embrace and use technology to drive forward improvements in patient care. We want artificial intelligence to be a tool at a physician’s disposal to help them do the best job possible. This means person-centred care, where technology helps both patients and doctors make the best possible decisions for their care.
On data, we want regulators, NHS England and NHS Digital to apply guidance, underlying principles and evaluation methods to technology. This is to ensure they remain safe, efficient and doctor-led as it is frontline staff who will be using it.
Analysis of data is crucial to the work of AI. For it to be effective, high-quality data is needed to make sure the conclusions drawn from AI are accurate and safe. But, patient data must be used transparently and sensitively.
We also want the data used to be broad and reflective of the general population, and ensure it is founded on robust evaluation and evidence.
AI is developing rapidly, and presents a lot of exciting opportunities. But it requires a lot of thought also, and guidelines to ensure it works in the best possible way for us.
Dr Andrew Goddard is the president-elect of the RCP. You can follow him on Twitter: @bodgoddard.
Read the RCP’s full position statement on AI, including recommendations.