Game changer
The headlines are all about artificial intelligence but what does it mean for the future of healthcare?
By Caroline Roberts
Illustrations by Eliot Wyatt
Game changer
The headlines are all about artificial intelligence but what does it mean for the future of healthcare?
By Caroline Roberts
Illustrations by Eliot Wyatt
Artificial intelligence, or AI, has the potential to revolutionise so many aspects of healthcare.
From analysing scans and finding new drugs, to creating personalised treatments, AI can improve patient outcomes. It’s not something we should be afraid of. The type of AI used in medicine is mostly to make predictions and process large amounts of data, to help doctors make better, faster decisions. This is different from AI models, like ChatGPT, which create new content and have raised concerns about misinformation and the risk of machines eventually going rogue.
Guiding surgeons
“Computer programmes and robotics are currently used to assist in joint replacement surgery, but they serve as guidance tools rather than intelligent systems,” says Mr Alex Chipperfield, Medical Director and Consultant Orthopaedic Surgeon at Benenden Hospital. “At the moment, AI is most useful for orthopaedics in the pre-surgical stages, where it can speed up the reporting of X-rays and scans.“
Over the years the National Joint Registry has gathered a lot of data about how well joint replacements perform. This data is now being used to create predictive models that can tailor implants to individual patients and predict the outcome of surgery based on their age, gender, weight and activity levels.
“It will give people a clear idea of what kind of improvements they would see in all aspects of their lives after surgery. It’s something that will come into the public domain over the next three to five years.”
Working together
The possibilities for AI in healthcare are wide-ranging, says Professor Mihaela van der Schaar, director of the Cambridge Centre for AI in Medicine at the University of Cambridge.
“Machines are good at some things and humans are good at others, and they can work together very well. Clinicians can’t hold huge amounts of information about disease trajectories in their heads but machine learning can help them understand trends, so it can be a fantastically supportive tool.
“For example, AI can help determine the rate of disease progression in individual patients. Some people with prostate cancer will die of old age but others have very aggressive cancer that needs aggressive intervention and AI can help differentiate between these.”
A tool to help the NHS
AI could also solve some practical NHS problems, she adds. “One thing I find very exciting is how we can use AI to help with resource management. At the beginning of the pandemic, our lab created a tool to help predict the need for resources such as ventilators and ICU beds. These tools could be used across the health service to streamline processes and help with long waiting lists.”
How can AI be integrated into an NHS with limited IT resources and many different programmes that are often unable to communicate with one another? The answer is cloud-based systems that clinicians can log into, says Professor van der Schaar. “All the systems we create are open source, or free to access on the web for NHS or research trusts. What it does cost is the time to learn how to use AI and capitalise on it.”
The Cambridge Centre runs a summer school and regular online education sessions for medical staff and researchers from all over the world. “As well as learning about AI, they’re telling us what they need us to work on. We’re trying to build an international ecosystem, which will be especially helpful in countries that lack doctors and resources.”
“AI is a very fast-moving sector with exponential growth expected over the next few years. With the right approach, we can make significant strides in improving patient care and outcomes.”
Protecting patient privacy
Patient privacy is one of the big concerns with AI. Anonymised patient data – like brain scans or photos of cancerous moles – helps AI learn and build a bank of information to help it make better decisions for future patients.
“Data management is a big topic, especially within a hospital environment where we’re dealing with sensitive patient data,” says Nick Lord, Head of Project Management Office at Benenden Hospital. “It’s vital that AI is used ethically and responsibly, in line with the hospital’s values. We’ve got to safeguard the data and ensure it’s not just secure but also accurate and reliable. It’s key that we do everything we can to operate in a safe way, and constantly monitor system performance to ensure it’s delivering expected results. Patient trust is vital.
“AI is a very fast-moving sector with exponential growth expected over the next few years. With the right approach, we can make significant strides in improving patient care and outcomes.”
Getting ahead of the game
Benenden Hospital is well-placed to become an early adopter, adds Mr Chipperfield. “Smaller institutions like Benenden Hospital are perfect for innovation as they’re much more contained and there’s a joined-up system already in place.
“AI is about innovation, about making things work better and making patients happier. I can’t see a negative; I can only be excited about what the future holds.”
Five ways AI might help in the future of healthcare
The National Institute for Health and Care Research (nihr.ac.uk) has identified some examples of how AI-based technology may be used in the future to help diagnosis and treatment:
1. Diagnosing heart disease Heart failure affects one in 100 people. A ‘smart’ stethoscope with additional sensors can help GPs diagnose and treat it earlier.
2. Cancer screening AI can be a useful tool in screening programmes. It can help determine whether abnormal findings on lung CT scans are cancerous. In breast cancer screening, it can almost halve the workload of radiologists.
3. Disease progression Predictive models can be used to assess whether patients with age-related macular degeneration in one eye will develop it in the other. They can also assess when those with inflammatory bowel disease are likely to suffer a flare-up.
4. Personalised treatments AI can help doctors choose the right combination of drugs for each individual cancer patient.
5. Managing resources Programmes can determine who needs to be seen in A&E and predict how many hospital beds will be needed in a few hours’ time.
