Trending Articles
Trending Topics
Trending
Peter Jones, an Advanced Clinical Practitioner in General Practice, explains why he think AI in healthcare is something we should be excited by
Listening to a recent PC Pro podcast (“Don’t sleep on the AI mattress“) has inspired me to commit some words to virtual paper in a bid to promote interest – dare I suggest excitement? – about AI in healthcare.
Discussion on AI has been done to death. Indeed, the illustrious PC Pro podcasters often preface discussion on AI with apologies that they’re about to cover it yet again. At the moment it does feel like AI is a lot of promise and little substance but I remain hopeful.
I concede that there are concerns; that AI lacks human intuition and common sense to see when a result or outcome is blatantly wrong, for example, not to mention issues around data protection. We have a way to go before AI is truly integrated into healthcare, but I also believe that with human checks in place it could be very powerful and make a significant difference in data and results interpretation.
Consider the role of AI in interpreting mammograms and using AI to detect breast cancer as a whole. The earlier that a cancer is detected the more likely a positive outcome: any technology that enables earlier detection must surely be a good thing.
But we also need to talk about the big issue: error in human interpretation of results. Whereas humans can apply reasoning and experience, we are also flawed creatures who can make mistakes.
Plus, medicine is seldom black and white. Trust me when I say that there isn’t a clinician in the land who hasn’t failed to spot a subtle fracture on an X-ray, because X-ray interpretation is an art form. Can AI reduce the instances of missed fractures? Very possibly.
One of the hardest parts of my role as Advanced Clinical Practitioner in General Practice is interpreting blood results. For a start “a” blood test usually includes several test groups, each providing several results.
Take possibly the most common blood test the “FBC” or Full Blood Count. This will yield a myriad of results, including number of red blood cells, size of the cells, numbers of different types of white blood cell, number of platelets and other things. And that is considered one test.
We may well ask for other tests at the same time and so we will receive a lot of results and accepted normal ranges back from the laboratory. All of these need to be interpreted with due consideration to the patient including their age, sex, race, health issues, medication and many other factors.
I can well imagine AI being able to analyse the results along with demographics and health information specific to the patient and produce potential diagnoses or recommendations for action or further investigation. Of course it would need a human brain to consider the output and whether it seems “right”, not to mention consider external factors.
A similar example is the ECG or electrocardiogram. This is when you have ten electrodes stuck to your chest and limbs in order to record the electrical activity of your heart.
Again this takes skill to read and understand. Modern ECG machines perform analysis and provide a crude interpretation, but this is based on one piece of data.
Now consider having AI interfacing between the machine and the patient’s records and so being able to consider the patient’s physicality (for example, obesity can affect the reading) description of symptoms and medication and provide much more valuable analysis.
These are specific examples but many more come to mind.
Robotic surgery for knee replacements is currently available in the private sector but not the NHS. This makes the surgery much more specific to the individual and provides better surgical outcome. I don’t know if AI is a part of this process but there must surely a place for AI in such procedures?
So far I’ve just applied AI to data but we can go further. With some types of dementia the person may experience change in colour perception, which affects their interaction with the world. People with Parkinson’s disease will often find themselves unable to walk from one floor pattern to another, instead shuffling in one place as if on a cliff edge.
Now consider the role of technologies such as the Apple Vision Pro and how they could use AI to interpret the environment and alter the colour or pattern of objects or the surroundings to enable the wearer to move freely and effectively.
We have hurdles to overcome and standards to put in place. AI needs to be applied with caution. But I genuinely believe that it has high potential for being a significant and positive aspect of healthcare in the future. It should be exciting.
