The Future of Artificial Intelligence AI in Diagnostics

In an interview with Qaiser Malik, Living Room Health’s Clinical Director of Radiology and a renowned musculoskeletal radiologist, we discuss the utilisation of AI in diagnostic imaging. The interview highlights how AI can alleviate administrative tasks in radiology, such as scan review, enabling doctors to dedicate more time to patient interaction, providing explanations and conducting thorough reporting.

How do you think the rise of AI will impact radiologists?

In terms of health care, there was a lot of hype when it first came about, and everyone thought that there wouldn’t be any need for certain types of jobs, particularly radiologists. People assumed that AI would read the scans and that would be it, but we quickly realised that that was a long way off. It can make us faster and help us make fewer mistakes by helping us prioritise what scans to read first. Instead of AI diagnosing a scan, it can say whether a scan is likely to be relatively normal, which will determine whether it is a priority or not. Now, there are lots of ethical considerations around that, but at the moment, no AI application is signing off on any kind of report. But this kind of system could help prioritise the huge NHS backlog of scans that need to be read. So, it’s going to make us faster, more efficient, and safer, but it’s not necessarily going to replace us. There is a saying in the radiology world that AI won’t replace radiologists, but it will replace radiologists who don’t know about AI.

Do you think that AI being used more often in medicine and healthcare will decrease human contact?

I think it will do the opposite. When we switched from paper to electronic records, it freed up more of our time with patients. It’s just progress, and we need to use the technology to our advantage. You have more patient contact and less time doing some of the menial tasks, so you can have more patient contact and more time to spend with the patient rather than grappling with the technology side of things.

So what would happen, and who would be accountable if AI is wrong?

That’s the million-dollar question. No one has figured it out yet. You can see the same issue with driverless cars, where society needs answers because the technology is here, and yes, it’s really exciting. It can make a big difference, but there is no real accountability yet. When a human makes a mistake, there is a process set up to deal with it, and we know human error is a thing. As a society, we’re not necessarily ready to accept that machines can also make mistakes.

Do you think it's possible to avoid machine bias in diagnostic imaging?

I think it’s all under research. Because AI is programmed by people and people make mistakes, there is a risk of bias, and there is the risk of building in bias. For example, if you show an algorithm many chest X-rays from London and then go and deploy the algorithm in Scotland, you may get a different output because you put bias into the algorithm. Because in London, you’ve got a diverse population with a moving population, loads of migration and lots of ethnic minorities. Whereas somewhere more rural might not have that same population and it might perform differently. So I think that kind of inherent issues need to be looked at, and we need to be aware of them.
I don’t know what the answer is. I think we need to do more and more work and keep testing, and we need more regulation from some of the regulatory bodies. So CQC, for example, or NHS England or these kinds of organisations need to really be involved because, at the moment, we don’t have answers to some of those questions.

Is there something you're most excited about or looking forward to within the future of medicine?

I think that doctors will have more time to focus on patients. Currently, we spend a lot of our time doing paperwork and administrative tasks. As a radiologist, for example, I spend a vast amount of my time vetting (saying whether a scan needs to be worked on and what needs to be done). Now, that could be completely automated using AI, which would allow me to use that time to explain the outcome of the result to the patient or do more reporting. And other clinicians focus a lot on checking blood and blood tests, looking for scan results, and many things they do during a consultation which don’t necessarily focus on interaction with the patient. So I think that’s where AI will have its real impact, giving back time to spend with patients.

If you would like to book an MSK consultation with Qaiser or book an MRI scan you can enquire here.

JOURNAL

>

Interviews