Regina JM Kamstra1, Tim J van den Broek1, Karleen F Pietersma1, L. Ijsselstijn2, Willem J. van den Brink1
1TNO, the Netherlands
2 RHMDC, the Netherlands
Background: Micro- and macrovascular complications of type 2 diabetes (T2D) often develop before clinical diagnosis. Current monitoring is episodic and largely glucose-focused. Wearable-derived parameters, such as heart rate, blood pressure, activity, and sleep in addition to glucose may provide insight into cardiometabolic health. This pilot study examined associations between non-invasive digital biomarkers and clinical variation related to complication burden.
Methods: An interim analysis was performed in 28 of 60 planned overweight adults (BMI 25–40 kg/m²) across the dysglycemic spectrum. Participants completed baseline clinical assessments and 10 days of continuous monitoring using a glucose monitor, Corsano Cardiowatch 287, and McRoberts DynaPort 7. Clinical variables were reduced by PCA, and digital biomarkers were correlated with clinical PC scores using Spearman analyses.
Results: Preliminary analysis identified two axes of cardiometabolic variation: a metabolic–insulin–renal stress axis (PC1), consistent with insulin resistance and early endothelial or renal changes, and a lipid-driven cardiovascular risk axis (PC2), reflecting atherogenic lipid profiles that were largely independent of glycemic status. For PC1, lower morning activity relative to later-day activity was associated with higher PC1 scores (ρ = −0.40, p = 0.035). Mean pulse pressure showed modest associations (ρ = 0.34, p = 0.077). CGM-derived metrics showed weak positive associations. Shorter sleep duration, lower sleep efficiency, and lower deep sleep proportion aligned with higher PC1 scores. For PC2, associations were generally weaker and not statistically significant. Shorter total sleep duration (ρ = −0.36, p = 0.057) and lower sleep efficiency were associated with lower PC2 scores, whereas deep sleep proportion showed a positive trend (ρ = 0.32, p = 0.097). Sedentary time and CGM time-in-range showed weak negative associations with PC2.
Discussion/Conclusion: These interim findings indicate that wearable-derived biomarkers, particularly activity timing and sleep measures, may capture cardiometabolic variation beyond glucose alone. Further analysis on a larger sample size in people with a broader range of T2D complications are planned.