Multimodal knowledge assortment in PROGRESS. Credit score: Nature Medication (2025). DOI: 10.1038/s41591-025-03849-7
To diagnose both sort 2 diabetes or pre-diabetes, clinicians sometimes depend on a lab worth often known as HbA1c. This check captures an individual’s common blood glucose ranges over the last few months. However HbA1c can’t predict who’s at highest threat of progressing from wholesome to prediabetic, or from prediabetic to full-blown diabetes.
Now, scientists at Scripps Analysis have found that synthetic intelligence can use a mix of different knowledge—together with real-time glucose ranges from wearable screens—to offer a extra nuanced view of diabetes threat.
The brand new mannequin, described in Nature Medication, makes use of steady glucose monitor (CGM) knowledge alongside intestine microbiome, eating regimen, bodily exercise and genetic info. It flags early indicators of diabetes threat that customary HbA1c exams might miss.
“We showed that two people with the same HbA1c score can have very different underlying risk profiles,” says co-senior writer Giorgio Quer, the director of synthetic intelligence and assistant professor of Digital Medication at Scripps Analysis. “By bringing in more data—how long glucose spikes take to resolve, what happens to glucose overnight, what the food intake is, and even what’s happening in the gut—we can start to tell who’s on a fast track to diabetes and who isn’t.”
“Ultimately, the goal of this work is to get a better understanding of what is driving diabetes progression and how we can intervene earlier in the clinic,” provides co-senior writer Ed Ramos, the senior director of digital scientific trials at Scripps Analysis.
Whereas some variation in blood sugar is totally regular—particularly after consuming—frequent or exaggerated glucose spikes is usually a signal that the physique is struggling to handle sugar successfully. In wholesome people, blood sugar sometimes rises and falls easily. However in folks vulnerable to diabetes, these spikes can change into sharper, extra frequent or slower to resolve, even earlier than routine lab exams like HbA1c choose up an issue. The brand new research exhibits that monitoring these day-to-day dynamics supplies a way more detailed view of an individual’s metabolic well being, and would possibly assist establish hassle earlier.
The findings are a results of a multi-year, digital analysis program referred to as the PRediction Of Glycemic RESponse Research (PROGRESS). The research used social media outreach to enroll greater than 1,000 folks from throughout the U.S. in a completely distant scientific trial. Members included folks with diagnoses of both pre-diabetes or diabetes, in addition to wholesome people.
Glucose spike definition. Credit score: Nature Medication (2025). DOI: 10.1038/s41591-025-03849-7
For ten days, they wore Dexcom G6 CGMs, tracked their meals and train, and despatched in samples of their blood, saliva and stool for testing. The researchers additionally had entry to individuals’ digital well being data, which included earlier lab values and diagnoses made by medical practitioners.
“This was a really pioneering effort in the remote clinical trial space,” says Ramos. “We had to design a study that participants could complete entirely on their own—from applying sensors to collecting and shipping biological samples—without ever visiting a clinic. That level of self-guided participation required a completely different kind of infrastructure than usual.”
Utilizing the information, the researchers educated an AI mannequin to differentiate folks with sort 2 diabetes from wholesome people.
One of many clearest indicators of diabetes threat that the researchers discovered was the time it took for a blood sugar spike to return to regular. In folks with sort 2 diabetes, it usually took 100 minutes or extra for blood sugar to lower after a spike, whereas more healthy people returned to baseline a lot sooner. The research additionally discovered that folks with a extra various intestine microbiome and better exercise degree tended to have higher glucose management, whereas the next resting coronary heart price was linked to diabetes.
Importantly, the AI mannequin did not simply detect threat in folks with already elevated HbA1c. When utilized to pre-diabetic people, it discovered that some seemed metabolically much like these with diabetes, whereas others resembled wholesome people, regardless of having related lab values. This degree of granularity might assist clinicians personalize therapy—specializing in life-style modifications or early therapies for sufferers with the best threat of illness development.
Whereas the present research was a snapshot in time, the researchers are persevering with to comply with individuals to see whether or not the mannequin’s predictions translate to real-world illness development. In addition they validated the mannequin utilizing a separate set of affected person knowledge from Israel, strengthening its potential for broader scientific use.
The workforce envisions future variations of the mannequin being utilized by clinicians, and even people utilizing CGMs at house, to evaluate metabolic threat and monitor how each day decisions have an effect on diabetes.
“Ultimately, this is about giving people more insight and control,” says Quer. “Diabetes doesn’t just appear one day—it builds slowly, and we now have the tools to detect it earlier and intervene smarter.”
Extra info:
Mattia Carletti et al, Multimodal AI correlates of glucose spikes in folks with regular glucose regulation, pre-diabetes and sort 2 diabetes, Nature Medication (2025). DOI: 10.1038/s41591-025-03849-7
Supplied by
The Scripps Analysis Institute
Quotation:
AI mannequin makes use of glucose spikes to disclose hidden diabetes threat earlier than signs seem (2025, July 31)
retrieved 31 July 2025
from https://medicalxpress.com/information/2025-07-ai-glucose-spikes-reveal-hidden.html
This doc is topic to copyright. Other than any truthful dealing for the aim of personal research or analysis, no
half could also be reproduced with out the written permission. The content material is supplied for info functions solely.
