It was only a matter of time, right? A 30-something patient presents with new pathology results saying she’d already asked AI to interpret them 🤖🤯 “Had a little chat with Chatty,” she tells me, “and already it’s flagging things my GP didn’t’!” Naturally, she’s tickled by this relatively new opportunity for what appears to be an independent second opinion, as a result of some simple tinkering on her keyboard. I watch her eyebrows rise to emphasise how impressed she is. She thinks she’s cracked the code, cut the cord of dependency to her doctor, or any other health professional for blood test interpretation. But is it true?
I was forced into being an early-adopter of AI in terms of large language models by my offspring, who argued that if I wasn’t on the front foot with these new technologies I would instead get crushed under its Big Foot 👾
Like every next gen, they gave me the, ‘look this is happening anyway so you may as well quit the denial’, talk
And because of that I’ve had a long lead-time to play around with these applications in a work setting and know how to play to their strengths, and accordingly the very short list of things they can be good for, but am always acutely aware of serious shortcomings, e.g. hallucinations, calculations, confirmation bias. Oh yes me and my team have found over time you can pretty much talk AI into anything…it is the jury and it loves to be led! So I’m long long long past the honeymoon period with these kinds of AI offerings but appreciate others aren’t. Like this patient with her pathology results and her rush to acquire an opinion from AI who shared the interpretation of her results with me, so I could check my own predictions about what it would be well and not so…and yeah they were all to be expected. Let’s just take one little example of that: her total bilirubin is consistently in the teens. Did chatGPT flag this as Gilbert’s syndrome? Of course not, as it’s technically ‘within range’. Does she have Gilbert’s Syndrome – absolutely.
But AI is incapable of really replacing us humans, as diagnosticians & practitioners because, as I heard someone say in an interview, ‘it effectively just parrots back the ‘average’ of all online information, most of which is of rather average quality to begin with!
So if AI assisted pathology interpretation is really just an echo chamber for what is the average of all online information: unquestioning adherence to reference ranges, where only results outside of this attract comment and even their interpretation is, at best, ‘average’ then I ain’t out of a job anytime soon and neither are you! A strength of this software is pattern recognition, essential in the realm of results interpretation, so that could be perceived as an added advantage. But helloooooooooooooooooo as holistic health care providers are we not all about pattern recognition?!
A favourite quote of a health care professional working in prevention and lifestyle medicine says:
“At the risk of seeming Luddite, the future of lifestyle medicine is humanistic rather than technological”
Gray et al., 2020
And any practitioner keen to embrace AI with the intention of saving themselves ‘on average 15mins of non-billable time per client reviewing pathology results’…Sorry, but this says straight up you’re not doing it right. Because if you were it’d be taking longer 😉, you’d know it is the single most satisfying element of patient work up that you never want to give up and you’d absolutely know, that AI can never replace real knowledge, rigorous diagnostic reasoning & appropriate taking into account of the very real individual these results belong to 👊🤓🎤
NB AI was not used to write this blog…it would have been much better if it had have though & also would have definitely saved me some time! 😂Now THAT is a good use for these AI
The primary objective of MasterCourse I is to realise the true value we can extract from the most commonly performed labs.
Accordingly, this training is appropriate for any health professional who considers standard medical blood tests as part of their patient assessment and work-up, including (but not limited to) naturopaths, nutritionists, herbalists, osteopaths, chiropractors, physiotherapists, midwives, nurse practitioners and doctors.
