Could Artificial Intelligence Revolutionize Diabetes Care?

By Corinna Santa Ana

3 min read

Sometimes, diabetes makes for strange bedfellows. Recently, I've been reading about how researchers are applying artificial intelligence (AI) to type 2 diabetes (T2D) diagnosis and care.

Against the backdrop of all the hoopla over generative AI systems (like ChatGPT), these 2 things might not work together at first glance. However, scientists are applying AI's analytical and predictive capabilities to predict who will most likely be diagnosed with diabetes and which complications they face the highest risk of developing.^1-3

Potential uses of artificial intelligence in diabetes care

Some people think this just might be a dramatic game changer when it comes to interrupting the seemingly inevitable path from prediabetes to a type 2 diabetes diagnosis and then on to further health complications and comorbidities later in life.^3,4

What if you could predict your odds of type 2 diabetes?

With AI, we might soon be able to identify a person's individual odds of developing type 2 diabetes and complications long before any signs or symptoms appear.

Currently, people's risk for developing type 2 diabetes is based on generalized risk factors like being overweight or having a family history. However, artificial intelligence can calculate a person's personal risk profile using an algorithm based on real-life health data and results collected from thousands of people over decades.³

While AI can help uncover these risks, it doesn't necessarily offer new treatments or resolutions. So, knowing your risk level doesn't guarantee there will be a way to resolve that risk entirely.

Applications of AI in type 2 diabetes care

Early diagnosis

The promise of applying AI to diabetes is that people will have a more accurate read earlier on of their likelihood of developing diabetes or complications.^3,4

It's thought that an earlier warning will give people and their healthcare providers extra time to respond in ways that reduce their risk and just maybe, if caught early enough, even prevent type 2 diabetes from developing at all.⁴

Preventative lifestyle changes

With AI, people may have the opportunity to apply treatments and health habits that are already recognized as beneficial. Maybe understanding their individual risk more clearly will make it easier to adopt new nutrition, exercise, and stress management habits.⁴

For some, making these lifestyle changes before a formal diagnosis could slow down and possibly even stop the progression of type 2 diabetes, resulting in more healthy years of life.

Personalized treatments

Healthcare providers will be able to offer highly personalized treatments long before complications develop. These treatments could include prescribing medications for their protective and preventive effects. Also, health screenings, like monitoring kidney function, can start before any indication that organ function has been damaged.⁴

Would you want to know?

Having the kind of early warning that comes from applying AI to type 2 diabetes could be instrumental in shifting our response to type 2 diabetes away from being reactive toward being preventative and proactive.

But by itself, it cannot guarantee that type 2 diabetes or the complications that come with it can be completely avoided. For some people, the best AI might offer is to delay the inevitable. And this only after making substantial lifestyle changes.

This begs the question: Would you want to know what your individual risk level is for developing type 2 diabetes?

Is knowledge enough to change health behaviors?

Would that knowledge be enough to motivate making lifestyle changes or adopting a different treatment approach? Even if these actions would only delay the inevitable?

If I went back in time to before my diabetes diagnosis, I'm not sure how I would answer these questions. They say knowledge is power. But if my response can only have a limited effect on my health, it's unclear how much power is actually in it.

What do you think about how applying AI to type 2 diabetes care might or might not change things for you?

This article represents the opinions, thoughts, and experiences of the author; none of this content has been paid for by any advertiser. The Type2Diabetes.com team does not recommend or endorse any products or treatments discussed herein. Learn more about how we maintain editorial integrity here.

Join the conversation

Please read our rules before commenting.

SweetEnough Member
I suppose AI couldn't do any worse than scientists in prediction. There is a problem, the same one faced by weather forecasters. Weather forecastesr make forecasts giving a percentage of likelihood of a weather condition happening on a given day. This percentage is based on the number of days with similar conditions or pre conditions where a given type of weather happened. The problem is that they get it wrong too often and sometimes badly wrong. Would any one of us want to be told that we are doomed to Diabetes which then may progress to an earlier death and then find out the prediction of that likelihood was wrong? How many of us would change our behaviour if we were told that we were very likely to get T2 Diabetes? T2 runs in my family but I didn't change my behaviour, I drank too much alcohol and ate too many carbs and didn't exercise enough. I'm sure I'm not the only one to behave like that. Heck, how many of us go on a low or very low carb diet to manage T2 Diabetes? I would guess very few. The only thing that might change our behaviours is relentless education with relentless emphasis on what might happen if we don't change our behaviours. Maybe a visit to hospital to see amputees or people on kidney dialysis or maybe talk to some people who lost their sight because of poorly managed diabetes. That's a few of the horrible complications that could happen to us. However I suspect the real enemy are those people in the medical industry who feel that diabetics mustn't feel marginalised and they talk about how managing diabetes is just eating healthily. Complete and utter rubbish as I'm sure some/most diabetics will attest to. These medical people say that telling us the truth about the seriousness of our condition will discourage us and we won't do anything, well some/many/most people don't do anything in any case. Tell people the unvarnished truth about diabetes and give them the tools which they can use to manage their condition. Give them personal responsibility and the power to manage that condition instead of treating them as passive victims "Doctors know best" and "Trust me I'm a doctor."
1. Lori.Foster Community Admin
 That kind of doctor is rare, indeed, <inline-mention username="SweetEnough" user-id="4506842"/>! I wish you didn't have to leave him behind. - Lori (Team Member)
2. Tula62 Member
 <inline-mention username="SweetEnough" user-id="4506842"/> a visit to the hospital to see amputees ect, that would not work for me I would be full of anxiety, and stress right out, things like that for me would make me go into diabetes distress, and send me back to psych ward, I have a hard time reading on here, it’s overwhelming, and when I was diagnosed I heard to much of that fear mongering and ended up with full blown panic attacks so absolutely NO for a lot of people.
michaelsdonohoe Member
The simple answer is that AI will affect primary care and chronic conditions. Since AI is based on learned knowledge, there are a few things to keep in mind. First, as with WebMD/Medscape & Healthline for example, these organizations follow the same processes as a pharma company or med device company. Basic Google, Bing, Chat GPT have a risk of picking up on non medically reviewed copy. Dr. Google can be dangerous. Second is that teaching processes including diabetes self management, using approved methodologies can be shared, but two people living with T2D can be very different. If someone is diagnosed with the condition today at age 55, it can be much different then my 27 years of living with it, even though I am 55. Individualized care is key, especially when you deal with the human condition. Thank you!
1. Lori.Foster Community Admin
 Those are such legitimate considerations, <inline-mention username="michaelsdonohoe" user-id="6723698"/>. It will be interesting to see what becomes of AI as a tool for the medical community, if anything comes of it at all. Thanks for joining the conversation. Warmly, Lori (Team Member)
2. lauren.mullin Community Admin
 <inline-mention username="michaelsdonohoe" user-id="6723698"/> Hi there, I just want to take the time to highlight what I feel to be the most important part of your post. That every person is different and individualized care is key. There is no one size fits all when it comes to health care, especially type 2 diabetes. All the best, Lauren (team member)
michaelsdonohoe Member
CopyL <inline-mention username="Corinna Santa Ana" user-id="3541819"/>. There are some PCP AI services now available, for a low rate of pay. In my opinion, and having worked in digital health promotions for pharmaceutical brands, I know that some of programs we created over the past twenty years (no AI) have worked. I believe there is an opportunity to bring health equity into lower served or improperly served communities. I plan on testing this over the next year, here in New York State, once I can find a partner I feel I can work with. AI could help on so many levels as a virtual coach, RD or trainer. Meal planning and recording in an easier way, integration with exercise (counting any movement) and so many other ways. I am sure it is out there, but I need to look more to be a help.
1. Corinna Santa Ana Moderator & Contributor
 I'm curious to see what develops. And I want to be hopeful. However, I realize how narrow a lot of medical research datasets are and how the very groups health equity targets are often un- or under-represented. 
2. lauren.mullin Community Admin
 <inline-mention username="Corinna Santa Ana" user-id="3541819"/> Thanks for sharing your views on AI as it relates to diabetes. You bring up an interesting point about under-represented groups. I wonder how AI will manage multiple groups, ages, genders, etc. when computing a response. This will be interesting to watch unfold. All the best, Lauren (team member)