Classification Accuracy of the Wrist-Worn GENEA Accelerometer During Structured Activity Bouts: A Cross-Validation Study

Purpose: The purpose of this study is to determine whether the left wrist cutpoints of Esliger et al., for the triaxial GENEA accelerometer, are accurate for predicting intensity categories during 14 different activities including; treadmill-based, home and office, and sport activities. Methods: 130 adults wore a GENEA accelerometer on their left wrist while performing various lifestyle activities. The Oxycon Mobile Portable Metabolic Unit was used to measure oxygen uptake during each activity. Statistical analysis used Spearman’s rank correlations to determine the relationship between measured and estimated intensity classifications. Cross tabulation tables were constructed to describe under or over estimation of misclassified activities, and one-way chi-squares were used to test whether the accuracy rate of each activity differed from 80%. Results: For all activities the GENEA accelerometer-based physical activity monitor explained 41.1% of the energy expenditure. The GENEA correctly classified 52.8% of observations when all activities were combined. Five of the 14 activities showed no statistical difference in physical activity intensity classification estimation when compared to 80% accuracy, with 1 activity (treadmill jogging) showing statistically greater accuracy than 80%. For the remainder of the activities, the GENEA showed less than 80% accuracy for predicting intensity. Conclusion: Cross-validation of the proposed GENEA left wrist cutpoints classified the majority of activities performed significantly below the accuracy rate of 80%. Researchers should be cautious when applying the Esliger et al. cutpoints to a different population and activities not tested by those investigators.