Can Wearables Make Diabetes Management Easier?

Continuous glucose monitoring (CGM) can help people with diabetes manage their disease and prevent long-term damage to the heart, eyes, kidneys, and nerves. But CGM systems are expensive and invasive, requiring repeated skin punctures or implanted sensors. 

The team at Duke’s BIG IDEAs Lab, led by Center for Computational and Digital Health Innovation Associate Director for Wearable Technology Jessilyn Dunn, PhD, wants to see if wearable sensors can reliably estimate post-meal glucose changes and replace CGM. 

If wearables can produce similar accuracy to CGM, people with diabetes could monitor their glucose more comfortably, with consistent monitoring, earlier detection of glucose spikes, and better-informed treatment decisions. For people with prediabetes or those who are at risk of metabolic disease, it could provide an accessible way to detect early warning signs and intervene before disease develops.

Thanks to pilot funding from the Center, BIG IDEAs lab members Lauren Lederer, MS, a PhD candidate, and Mahmud Islam, PhD, a postdoctoral researcher, are collecting initial feasibility data and conducting early analyses.  

“This study builds on prior work in the lab showing that wearable data combined with contextual information can classify glucose states with promising accuracy,” Dr. Dunn says. “This pilot removes reliance on diet logging and instead explores whether advanced optical and multimodal sensors can recover that signal directly from physiology.”

If the pilot demonstrates promising signals, the next steps include:

1. Expanding the study to include individuals with prediabetes and diabetes.
2. Refining and optimizing machine learning models for real-time glucose estimation.
3. Exploring device miniaturization and integration pathways to enable scalable, user-friendly deployment.

In the long term, the goal is to move from proof-of-concept modeling toward clinically validated, non-invasive glucose monitoring systems.

Exploring the Potential of Wearables

Modern wearables and non-invasive optical and radar-based measures can capture rich physiological data. With their deep understanding of the capabilities of wearables, the team asked:

Could subtle physiological responses to glucose intake, like changes in cardiovascular signals, skin conductance, and optical biomarkers, be used to estimate blood glucose levels?

The study uses several research-grade and commercially available wearable sensors to capture complementary physiological signals related to glucose regulation. Together, these modalities produce a synchronized, multimodal dataset that enables investigation of how glucose fluctuations are reflected across multiple physiological systems.

The team will use the Stelo Dexcom continuous glucose monitor to provide a high-frequency reference for interstitial glucose levels, supplemented by periodic capillary finger-prick blood glucose measurements that serve as the gold standard labels when performing analysis and building models. 

From Hypothesis to Human Study

The team has completed study design, device integration planning, and internal protocol testing. The infrastructure for data collection is now in place. The next phase is active participant recruitment and pilot data collection.

“Our goal is to generate a high-quality, synchronized dataset of non-invasive sensor signals and gold-standard glucose measurements in a controlled lab environment,” Ms. Lederer says. “Success means demonstrating that these signals are strong and consistent enough to justify scaling the study beyond the lab and into broader, real-world populations.”

For more information on the work happening at BIG IDEAs Lab, visit their site.