Hot topic: Where is AI having the biggest impact in healthcare right now?

The big buzz around AI in healthcare is the potential it holds to identify and even create new drugs. “AI drug discovery” is the great new hope.

And – despite critics awkwardly pointing out that no AI-generated drugs have yet been approved by any regulator, and vanishing few appear to be close to that finishing line – it seems likely this gigantic project will deliver in the end.

But that end might be some way off. Furthermore, human biology is notoriously complex and liable to throw a spanner in the works of even the cleverest AI black box. There’s a reason why a good number of blockbuster drugs were esoteric discoveries. Viagra, which started life as a cardiac drug, is a classic case in point. When it comes to how experimental drugs will interact ‘IRL’ (in real life), very often we just don’t have a clue.

So, if the promise of AI drug discovery could take many years to fulfil, which shorter-term wins are possible?

Two applications are currently standing out. One: using AI (or more accurately machine learning) to tell if a patient’s scan shows disease or not. Two: using AI (or more accurately large language models, although they increasingly use generative AI as well) to automatically transcribe a doctor’s conversation with a patient, make structured notes, and put all the relevant data in the right boxes.

These might not be on the same level as the lofty ambitions we have for AI, of helping us cure cancer or stopping Alzheimer’s in its tracks.

Nonetheless, their significance should not be underestimated – because they could both help people working in hospitals and surgeries become much more productive.

Study after study is showing that programmes designed to learn from thousands of scans (including modalities like X-rays, ultrasound, CTs or MRIs) are just as accurate, if not more accurate, than experienced human experts trained to spot things like eye disease or breast cancer. But the programmes are, unsurprisingly, much faster.

Should radiologists, who review scans, be worried for their jobs? Probably not, because that’s just one part of what they do. In addition, there’s loads of work to go around: a major bottleneck when it comes to providing healthcare is manpower. And with ageing populations, that workload is only going to increase.

Similarly, ‘ambient scribes’, as they are known, are today enabling doctors and other clinicians to become more efficient, automating a time-consuming and tedious aspect of their work which is little more than data entry. This is freeing-up medics to see more patients. Given how long it can take to see a doctor, that’s a big win.

The importance of these scribe tools is being recognised by the market, with several companies such as Epic AI, Ambience Healthcare and Abridge now valued at over $1 billion. Abridge’s value doubled in just four months to $5.3bn, TechCrunch reported in June.

So, while we should certainly have our ‘eyes on the prize’ of successful AI drug discovery, it is by no means the only game in town. There are other AI technology games worth playing too.