University of Michigan Study: Many Smartwatch Health Readings Are Algorithmic Estimates, Not Direct Measurements

Your smartwatch's numbers may not mean what you think they do. A September 2026 study from the University of Michigan tested 10 commercial wearables against clinical-grade equipment in 45 adults and found significant accuracy gaps — particularly during movement. The findings arrive as wearable makers push deeper into health monitoring with features like blood pressure trends and sleep apnea detection.

Key Findings

• Best accuracy: Fitbit Charge 6 (4.5 BPM median error) and Google Pixel Watch 2 (4.9 BPM) performed closest to ECG measurements
• Wider errors: Polar Ignite 3, Polar Pacer, Oura Ring Gen 3, and Fitbit Inspire 3 showed median errors of 9.5–14.3 BPM
• Movement matters: Intermittent walking increased errors across all devices, suggesting resting readings deserve more confidence
• Sleep staging: Separately, a meta-analysis found some devices achieved only ~50% accuracy at correctly detecting sleep stages vs. clinical polysomnography

What This Means for Buyers

Wearable health data is best used for trend tracking, not diagnosis. Resting heart rate, activity duration, and step counts remain reliable. Sleep stages, blood oxygen, and stress scores are useful directional indicators but shouldn't replace clinical testing. Compare accuracy-focused wearables in our smartwatch guide and smart ring guide.