The new frontier of diabetes care

My husband is a secondary school music teacher and it’s been fascinating hearing educators’ thoughts on using artificial intelligence (AI) tools in the classroom. I’ve generally heard two schools of thought about AI, such as ChatGPT. Some teachers identify how useful it can be and are excited about the myriad ways it can enhance education and support students’ learning in new and innovative ways; others see it as pure evil, signalling the end of education as we know it and will be exploited by pesky students. Thankfully, my husband is very firmly in the first group!

My daughter is a freshly minted university student, and of course, AI is front and centre of university faculty minds, with students receiving warnings about misusing the technology, and how assessors would be on the lookout for any assignments turned in that appear more ChatGPT and less CramGPT-AllNight. But again, it’s interested to hear about how some of her lecturers and tutors are using the technology to develop novel and engaging ways to learn. And students are finding ways to use AI tools to assess the work they are doing and pinpoint shortcomings to allow them to go back in and work on specific areas that might need attention. 

But where are we in diabetes? What are we thinking about how AI could be useful in diabetes education and support? The ADA’s Diabetes Care journal weighed into the issue with this article published recently. The article provided a generally balanced view of what AI could do when it comes to providing diabetes education. 

Some of the pitfalls noted include: inaccuracy in the information provided, (and limited updates on new evidence and developments) and the potential for presenting false information convincingly, leading to safety concerns due to factual inaccuracies. There is often a lack of nuance in differentiating insulin types and variation in blood glucose units. Obviously, it has a reliance on a general information database, not specific medical knowledge. There was also a comment about there being inconsideration for the needs and wishes of individuals (but then, I’m pretty sure some HCPs I’ve had to deal with could be accused of the same thing). The reliance at the end of almost every response to ‘consult your healthcare team’ is a helpful if people have access to such a team but could be problematic for those who don’t. The downsides it listed were fair, although I suspect most could be easily overcome.

And the advantages that were highlighted were also spot on, and included: increased access to information, and the provision of systematic, concise, well-organised responses, with clarity around any jargon used. The ability to augment basic levels of education and offer more detail if requested by the user means that people with diabetes can customise their education to suit their level of understanding, needs and interests and regional and cultural contexts, rather than be at the mercy of generic education models and information that is often all that is offered. Of course the fact that education can be on-demand – when and where people with diabetes want it – is hugely convenient! These certainly are all valuable and could address many of the frustrations experienced by people with diabetes when it comes to seeking and receiving diabetes education.

But there was one glaring omission in the advantages list. ChatGPT comes with neutrality and with that neutrality comes a lack of prejudice about diabetes and people with diabetes. It doesn’t judge and it doesn’t blame the person with diabetes either. At no time is ChatGPT going to accuse the person seeking education that they are lying about their glucose levels (yes, that happened to me), or imply they are in denial about their diabetes (yes, so did that), or suggest the person is simply not trying hard enough (yes, and that) or isn’t caring about their health (yep – on more than one occasion). AI bots in diabetes care, and health more widely, can be seen to create a supportive space where people can seek guidance without feeling self-conscious.

The fact that there is no emotion in an AI bot means that they come without the baggage of preconceived ideas and previous experience. And they come without human interaction, and yes, I realise that this is in direct contradiction with my frequent calls for HCPs to be more human and person-centred. 

Before any diabetes educators come at me for trying to do them out of a job, please know that’s absolutely not what this post is about. Rather, I’m trying to highlight what it is that AI offers that could overcome some of the barriers (and those barriers can be mighty, mighty difficult to overcome!) that are experienced by people with diabetes when it comes to diabetes education.

I don’t think it needs to be said that this is a conversation that is going to be had more and more. AI isn’t in the scary future of diabetes education; it’s here and now, and smart CDEs will be the ones who work out how it can be used to enhance their work and be added as an additional tool to support people with diabetes. In fact, I would be looking for health professionals who are actively embracing this sort of tech rather than expressing scepticism about it. 

Because here’s the thing: AI is going to put more control and power into the hands of people with diabetes. It’s already doing that and it’s only going to increase as AI models get smarter and can learn and adapt to each user, faster and more accurately. And this is going to mean less reliance on HCPs for aspects of diabetes care. But this is a great thing, in exactly the same way that diabetes tech such as home glucose monitoring, insulin pumps. CGM and automated insulin delivery have delivered similar outcomes. 

The way diabetes care is evolving is at a crossroads. Better diabetes-specific technology, better drugs, ever-growing peer-led innovation, learning and support, all augmented by AI. The potential to create opportunities to better support people with diabetes and enhance our care is huge. I’m not for a moment suggesting that AI is a substitute for human interaction, but it undoubtedly has benefits over traditional healthcare approaches and offers new dimensions of support, in ways that can be customised and more relevant for people with diabetes. And if, at the foundation of all this, is the opportunity to remove stigma and help people with diabetes feel less blamed and shamed for their condition, that is a massive win!