Use of accelerometry to predict energy expenditure in military tasks
Horner, F., 2012. Use of accelerometry to predict energy expenditure in military tasks. Thesis (Doctor of Philosophy (PhD)). University of Bath.
Related documents:This repository does not currently have the full-text of this item.
You may be able to access a copy if URLs are provided below.
The overarching aim of this thesis was to enhance the prediction of physical activity energy expenditure (PAEE) in military personnel; specifically, improving accuracy and minimising obtrusiveness. The first experimental chapter provided a thorough assessment of the reliability and validity of the 3DNX accelerometer. Within unit reliability (CVintra) physical activity counts (PAC) was 0.0-8.9% in all axes in a mechanical setting. Between unit reliability (CVinter) did not exceed 4.5%.The relationship between PAC and acceleration was r2 = 0.99 and standard error of the estimate (SEE) of 6 counts∙5s-1. During treadmill exercise, the relationship between and PAC was linear (walking, r2 =0.65, SEE = 1.42 ml·kg-1·min-1; running, r2 =0.62, SEE = 3.63 ml·kg-1·min-1). 3DNX PAC output was valid and reliable when subjected to a physiologically relevant range of mechanically generated accelerations and yielded a linear relationship with during treadmill walking and running. Chapter 7 investigated the effect of anatomical placement on PAC in order to find the most suitable wear location. Hip and back placements returned similar reliability (CVintra = 3.0% and 2.8% respectively). Hip PAC were higher (p < 0.01) for walking with no differences observed for running. Indices of adiposity were related to hip PAC. Regression analysis revealed hip and back PAC as significant predictors of . Back PAC was the least variable placement. Supraspinale skinfold thickness explained 15% additional variance in to PAC and reduced SEE. In Chapter 8, three available devices were compared to doubly labelled water (DLW) for the prediction of free living PAEE using a user-oriented approach. All devices underestimated PAEE. Actiheart-derived PAEE was not different from DLW. However, the wide absolute limits of agreement (LoA) indicated large individual error which was attributed to the use of group rather than individual calibration. 3DNX and GT3X PAEE predictions were different from DLW however LoA were narrower indicating the possibility of applying a correction factor in future. Chapter 9 was an amalgamation of ten independent cohorts in an attempt to produce a military-specific multivariate model for the prediction of energy expenditure (EE). Stringent data reduction techniques were applied to a highly compliant dataset. Allometric models showed PAC, height and body mass were related to total energy expenditure (TEE) (p < 0.01). For models predicting TEE, PAC explained 4 % of the variance. For models predicting PAEE, PAC accounted for 6 % of the variance. The small amount of variance explained by PAC was likely due to the inability of accelerometers to detect EE as a result of day-to-day military activities such as load carriage. Such small portions of explained variance indicate that traditional accelerometry techniques are inadequate for use in military populations. In Chapter 10, an alternative approach to characterising military-specific activities was explored due to the minor contribution of PAC to PAEE prediction in Chapter 9. Accelerometer raw signal (100 Hz) was used to develop a classification model which aimed to discriminate load carriage (LC) from unloaded ambulation during an occupationally relevant protocol (2-hours, 6.4km∙hr-1, 25kg load). Fast Fourier transformation showed differences in the frequency distribution of the signal between conditions; caused by differences in gait parameters. Load carriage was detected in 97.2% of 1-minute samples with reduced classification accuracy during the last 30 minutes. Fatigue was suggested as a cause of misclassification; indicated by an upwards drift in and RPE across time. In conclusion, accelerometer PAC is a weak contributor to the prediction of energy expenditure in military populations. Accuracy could be improved by detection of load bearing activities which is feasible given the advancement in technology and analysis techniques. New technologies such as optical interferometrics could be integrated into existing military equipment to detect heartbeat and respiration; providing data regarding the physiological strain of training and operations.
|Item Type||Thesis (Doctor of Philosophy (PhD))|
|Departments||Faculty of Humanities & Social Sciences > Health|
|Publisher Statement||UnivBath_PhD_2012_FE_Horner.pdf: © The Author|
Actions (login required)