AI model uses glucose spikes to reveal hidden diabetes risk before symptoms appear

To diagnose both sort 2 diabetes or pre-diabetes, clinicians usually depend on a lab worth often called 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 danger 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 information—together with real-time glucose ranges from wearable screens—to offer a extra nuanced view of diabetes danger.

The brand new mannequin, described in Nature Drugs, makes use of steady glucose monitor (CGM) information alongside intestine microbiome, weight-reduction plan, bodily exercise and genetic info. It flags early indicators of diabetes danger that commonplace HbA1c checks might miss.

“We confirmed that two individuals with the identical HbA1c rating can have very totally different underlying danger profiles,” says co-senior writer Giorgio Quer, the director of synthetic intelligence and assistant professor of Digital Drugs at Scripps Analysis. “By bringing in additional information—how lengthy glucose spikes take to resolve, what occurs to glucose in a single day, what the meals consumption is, and even what’s occurring within the intestine—we are able to begin to inform who’s on a quick monitor to diabetes and who is not.”

“Finally, the purpose of this work is to get a greater understanding of what’s driving diabetes development and the way we are able to intervene earlier within 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 usually rises and falls easily. However in individuals prone to diabetes, these spikes can develop into sharper, extra frequent or slower to resolve, even earlier than routine lab checks like HbA1c decide up an issue. The brand new research reveals that monitoring these day-to-day dynamics gives a way more detailed view of an individual’s metabolic well being, and may 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 individuals from throughout the U.S. in a totally distant scientific trial. Contributors included individuals with diagnoses of both pre-diabetes or diabetes, in addition to wholesome people.

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 members’ digital well being data, which included earlier lab values and diagnoses made by medical practitioners.

“This was a very pioneering effort within the distant scientific trial area,” says Ramos. “We needed to design a research that members may full solely on their very own—from making use of sensors to gathering and transport organic samples—with out ever visiting a clinic. That stage of self-guided participation required a totally totally different type of infrastructure than common.”

Utilizing the information, the researchers skilled an AI mannequin to differentiate individuals with sort 2 diabetes from wholesome people.

One of many clearest alerts of diabetes danger that the researchers discovered was the time it took for a blood sugar spike to return to regular. In individuals 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 stage tended to have higher glucose management, whereas the next resting coronary heart fee was linked to diabetes.

Importantly, the AI mannequin did not simply detect danger in individuals with already elevated HbA1c. When utilized to pre-diabetic people, it discovered that some appeared metabolically just like these with diabetes, whereas others resembled wholesome people, regardless of having related lab values. This stage of granularity may assist clinicians personalize therapy—specializing in life-style modifications or early therapies for sufferers with the best danger of illness development.

Whereas the present research was a snapshot in time, the researchers are persevering with to observe members to see whether or not the mannequin’s predictions translate to real-world illness development. Additionally they validated the mannequin utilizing a separate set of affected person information from Israel, strengthening its potential for broader scientific use.

The staff envisions future variations of the mannequin being utilized by clinicians, and even people utilizing CGMs at house, to evaluate metabolic danger and monitor how every day selections have an effect on diabetes.

“Finally, that is about giving individuals extra perception and management,” says Quer. “Diabetes would not simply seem at some point—it builds slowly, and we now have the instruments to detect it earlier and intervene smarter.”