A research team at Google has developed an AI model that predicts insulin resistance with up to 88% accuracy by analyzing smartwatch data alongside blood tests. This breakthrough highlights the growing connection between Smartwatch and Insulin Resistance, offering a powerful tool for early detection of a key precursor to type 2 diabetes.
Researchers Combine Wearable Data With Blood Tests
MOUNTAIN VIEW, Calif. — A team led by Ahmed Metwally at Google Research has developed an artificial intelligence model that detects insulin resistance by combining smartwatch data with standard blood test results. Published on March 16 in Nature, the study underscores the growing link between Smartwatch and Insulin Resistance, pointing to new possibilities for early detection of metabolic health risks.
The model analyzes metrics such as heart rate, sleep patterns, and daily activity collected from wearable devices. Researchers say the approach could identify early warning signs of metabolic disorders before symptoms appear.
Insulin resistance occurs when the body’s cells do not respond properly to insulin, causing blood sugar levels to rise over time. If untreated, it can lead to type 2 diabetes.
“Early detection of insulin resistance allows for slowing or even reversing progression to diabetes through lifestyle changes alone,” the research team says in a statement.
Study Tracks 1,165 Adults Using Smartwatches
The study includes 1,165 adults in the United States who wear commercially available smartwatches, including Fitbit and Google Pixel Watch devices, for up to three months.
Participants’ wearable data, including heart rate, heart rate variability, step count, and sleep duration. This is combined with laboratory results such as blood glucose, lipid levels, and glycated hemoglobin, or HbA1c.
Using both data sets, the AI model predicts insulin resistance with about 80% accuracy. When researchers apply an additional system trained on 40 million hours of smartwatch sensor data, accuracy rises to 88%.
The model correctly identifies 76% of individuals with insulin resistance, even when their blood sugar levels remain within the normal range.
Researchers note that traditional detection methods require hospital visits and fasting insulin tests, which can limit early diagnosis.
Findings Highlight Broader Health Implications
The study finds that individuals with higher insulin resistance often show faster resting heart rates, lower variability between heartbeats, and reduced daily step counts. These insights strengthen the connection between Smartwatch and Insulin Resistance, demonstrating how wearable data can reveal early markers of metabolic health.
While 45% of obese participants show insulin resistance, researchers also identify the condition in 7% of individuals with normal weight, underscoring the limits of weight-based screening alone.
An estimated 20% to 40% of adults may have insulin resistance, though it often goes undiagnosed in early stages because blood sugar levels can appear normal.
The team also develops a conversational AI tool that delivers personalized health advice based on user data. In an evaluation by five endocrinologists, the system scores higher than existing AI tools for completeness, reliability, and personalization.
“As smartwatches become more widespread, this model could help manage the metabolic health of millions,” the researchers say.
Experts say broader adoption could support preventive care by encouraging lifestyle changes such as improved diet, regular exercise, and weight management before disease onset.
Visit Healthcare 360 Magazine For The Most Recent Information.
