Asking about AI: useful or useless for healthcare?

Here we are, October already. It feels like it was only five minutes ago I was wishing everyone a lovely summer break. But life goes on – sometimes too quickly! – and before you know it, we’ll be looking December in the eye, with 2024 lurking right around the corner.

I think one of the reasons the year has passed so quickly, is because of how busy things have been in the healthcare industry. Patients, research, new data protection innovations and data – it’s enough to keep all of us occupied. But life is sometimes at its best when we’re busy. We rise to new challenges and develop new skills. And, of course, new seasons of the year mean new discoveries just waiting for us to explore.

And in the world of healthcare, AI is still the thing that keeps on giving us all things ‘new’.

What’s new then?

We’ve spoken before about the pitfalls and potentials of AI in the healthcare and research industry, and I’ll be honest: I’m still a little bit on the fence about the pros outweighing the cons. However, something that has got me really quite excited, is the announcement that Google have added new AI-powered capabilities to their Vertex platform for health and science.[i]

The concept is wonderful: the ability for researchers and practitioners to ask relevant medical questions about a patient, and receive factually correct information in real time, based on the patient’s own medical history and published scientific data. It could make things quick, effective, and much easier for clinicians than an endless scroll back through (often unreliable) records.

Microsoft are doing something similar[ii]. Their Azure AI services are taking aim at patient healthcare data (records – written or typed; images; clinical letters and prescriptions) and starting to transform them into easy, reliable data that can be streamlined for clinicians reviewing patients.

More widely, with a critical number of staff vacancies, the NHS itself is now adopting more and more use of AI. From the analysis of patient imaging, to remote health monitoring for patients[iii], the inroads into new and innovative patient care are fast-moving and fascinating. Simple clinical tasks are now being performed by artificial intelligence, freeing up clinicians and administrators alike to focus on more complex issues.

So is it time to subscribe to AI?

The truth of the matter is that AI really does have its good points. The ability to access live data, to assist with diagnoses and the potential for staying up to date with patient condition and clinical research means there are huge benefits to using AI for more than just the (fantastic) discoveries we’ve spoken about before. Its use in a wider clinical setting is a definite help, and as it continues to grow and evolve, I suspect our reliance on it in the healthcare industry will grow too.

But.

There are still issues, and they are ones I wrestle with. Whilst in the UK the Information Commissioner’s Office (ICO) has some well-developed guidelines around the use of AI[iv] and the safeguards that must be put place to ensure data is protected appropriately, there is still the fact that AI is ever-evolving. Data protection guidelines and laws will have to continue to evolve with it, or run the risk of sensitive medical information being leaked, or used for incorrect purposes.

And then there’s the issue of consent. The use of patient information can be a power for good, but how many patients truly understand when their information is being viewed by a consented-to clinician, or an unknown AI? Understanding may be limited, and there isn’t time for already-busy staff to begin to explain where patient information (anonymised or not) may end up. There are also concerns around bias, lack of transparency and safety.

Still, as I’ve said before: AI can be a powerfully positive force. The negatives need reviewing carefully – and AI’s work should be consistently monitored to ensure accuracy – but there is much to celebrate with these latest developments. We will always need clinicians. Researchers. Administrators and support staff. They get the hard, sometimes complex, sometimes ethical decisions right, where AI understandably can’t. But to have easy access to appropriate information – to be able to speed up their own work safely and effectively and reduce the stressors of their workloads – well that’s invaluable.

Finally for me, it always comes back to patient care. What is in a patient’s best interest? What is going to help them? A clinician or researcher with a wealth of complete and reliable information at their fingertips, able to give a supported, accurate diagnosis can do that. AI can support them.

The year will continue to speed along, and one of the new challenges, changes and innovations we need to meet as healthcare providers is the hard fact that AI is here to stay.

And maybe as we move towards Autumn it’s time to slow down and reflect that perhaps that’s not such a bad thing after all.

[i] Hagan, J., Google debuts new Vertex AI search capabilities for healthcare providers, 09.10.23

[ii] Taylor, A., Microsoft introduces new data and AI solutions to help healthcare organizations…, 10.10.23

[iii] NHS England, Artificial Intelligence, 10.02.23

[iv] ICO, Artificial Intelligence, 10.10.23