LET’S play a word association game.
Are you thinking what I’m thinking?
If so, you‘re thinking about drug discovery.
It is a very hot topic in pharma right now. Billions are being poured into efforts to use artificial intelligence (AI) to spot new drugs.
And for good reason. ‘Traditional’ drug discovery methods are often slow and eye-wateringly expensive. Needles and haystacks come to mind.
If we can use AI to circumvent the process, to quickly identify molecules that are most likely to treat a disease successfully, so much the better.
What’s more, some companies are going a step further and actually creating drug candidates from scratch using AI.
But it would be a mistake to think this is the only AI game in town.
Earlier this month, attendees at The AI Summit at Tobacco Dock near Tower Bridge in London heard about a range of other ways in which AI is being explored to improve healthcare.
A rapt audience listened to computer scientist Dr Lucrezia Cester speak about how she has developed a system to check for cardiovascular disease by scanning the patient’s neck from several feet away, using a device that combines a low-powered laser and a high-speed camera.
Sounds like science fiction, doesn’t it?
The device gathers data on the patient’s blood flow, which is analysed using a machine learning programme to determine if the individual has a healthy heart or not.
The project, named Light-Hearted AI, has NHS backing and is being rolled out for clinical use in Dorset.
It holds out the hope of GPs being able to screen patients for signs of heart disease in local surgeries, rather than having to refer them on to hospital for expensive tests like ECGs – although there is a lot of validation to go through before that becomes a reality.
Light-Hearted AI is an example of another top trend, using AI / machine learning to spot patterns in medical data that humans find hard to do ‘by hand’.
AI pattern recognition is being used in a myriad of medical fields, from detecting eye diseases, to cancer, to malaria.
However, while academic papers in this area are piling up fast, Dr Cester said there are still only a handful of cases in which AI is actually being used in a clinical setting.
The AI Summit also heard how the technology could be used to help hospitals better manage their inventories – potentially saving them money (and the environment from unnecessary waste) by smarter ordering.
Dr Costas Fantis, of Encode Health, explained that the idea is to use ‘big data’ to predict what operations or procedures will be undertaken in the future, to optimise purchasing plans – particularly of products that only have a limited shelf life for safety reasons.
NHS trusts are also looking at using AI to analyse data sets to better predict which A&Es will be busy at what time, and hence where to send ambulances. But there is understandable hesitation in deploying such technology in potentially life or death situations, said Simon Mortimore of South Central Ambulance Service NHS Foundation Trust, in case the algorithm gets it ‘wrong’. Decisions taken by new ‘black box’ AI technology are likely to be scrutinised more closely than those made by human minds – even though we are far from infallible ourselves.
These examples barely scratch the surface of how AI-related technological developments might change healthcare in the coming years. We could talk about how ChatGPT and other ‘large language models’ – which trawl the internet for information – could be used to diagnose diseases, or at least triage patients. Or how wearables could be used to predict when we are about to have a heart attack. Or how AI can be utilised to better schedule medical appointments. Or manage payment systems.
So, no, AI drug development is not the only game in town.
Saying that, it’s a pretty important game – or investors wouldn’t be betting the farm on it.