Smartwatch Calorie Burn Estimates Show 15-25% Median Error Rising With Body Fat Percentage in PLOS One Cycling Trial

The Florida International University team fitted 58 adults with Apple Watch Series 8, Garmin Forerunner 955, Samsung Galaxy Watch5, and Fitbit Sense 2 while they alternated moderate and vigorous bouts on a recumbent bike. A metabolic cart provided the criterion measure of energy expenditure. All devices showed substantial absolute error; three systematically overestimated calories. The magnitude of overestimation grew reliably with higher body-fat percentage, a pattern the watches' algorithms did not correct despite user-supplied weight and demographic data.

Wearable algorithms combine accelerometer and photoplethysmography signals with static inputs, yet they omit direct measures of body composition, movement economy, and metabolic flexibility that differ across adiposity levels. Prior observational work in JAMA Cardiology and a 2023 meta-analysis in Sports Medicine similarly documented larger errors in individuals with BMI >30, suggesting the bias is reproducible rather than device-specific. Public-health implications are direct: users with higher body fat who rely on these readings for weight-management decisions may receive systematically inflated feedback.

Regulatory bodies have not required accuracy benchmarks stratified by body composition. Future validation studies must recruit diverse adiposity ranges, report error by body-fat deciles, and test whether firmware updates that incorporate impedance or user-reported body-fat data measurably reduce bias. Without such stratification, wearable-derived energy estimates remain unsuitable for individualized clinical counseling.