Doctoral Dissertations

Date of Award


Degree Type


Degree Name

Doctor of Philosophy



Major Professor

David R. Bassett, Jr.

Committee Members

Edward T. Howley, Dixie L. Thompson, James W. Bailey


This dissertation was designed to examine the validity of heart rate (HR) and motion sensors for estimating energy expenditure (EE) during activities ranging from sedentary behaviors to vigorous exercise. A secondary purpose was to devise new ways to improve on current methods of estimating EE. Specific aims of the dissertation were: (1) to examine the use of pedometers to measure steps taken, distance traveled, and EE during treadmill walking at various speeds; (2) Examine the use of a Polar HR monitor to estimate EE during treadmill running, stationary cycling, and rowing; (3) compare the current Actigraph regression equations (relating counts·min-1 to EE) against three newer devices (Actiheart, Actical, and AMP-331) during sedentary, light, moderate, and vigorous intensity activities; and (4) development of a new 2-regression model to estimate EE using the Actigraph accelerometer.

For the first aim, 10 participants performed treadmill walking for five minutes at five speeds while wearing two pedometers of different brands (10 pedometer brands were tested) on the right and left hip. Simultaneously oxygen consumption (VO2) was measured and actual steps were counted using a hand tally counter. Six of the 10 pedometers were within ± 3% of actual steps at 80 m·min-1 and faster. Most pedometers were within ± 10% of actual distance at 80 m·min-1, but they overestimate distance at slower speeds, and underestimate distance at faster speeds. Most pedometers gave estimates of gross EE within ± 30% of measured EE across all speeds. In general, pedometers are most accurate for assessing steps, less accurate for assessing distance, and even less accurate for assessing kcals.

In the second aim, 10 males and 10 females performed a maximal treadmill test. On a separate day they performed treadmill, cycle, and rowing exercise for 10 minutes at three different intensities. During each trial EE was estimated using two Polar S410 HR monitors (one with predicted VO2max and HRmax (PHRM) and one with actual VO2max and HRmax (AHRM), input into the watch). Simultaneously, EE was measured by indirect calorimetry (IC). For males there were no differences among the mean values of EE for the AHRM, PHRM and IC for any exercise mode (P ≥ 0.05). In females, the AHRM significantly improved the estimate of EE compared to the PHRM (P < 0.05), but it still overestimated mean EE on the treadmill and cycle (P < 0.05). The Polar S410 HR monitor provides the best estimate of EE when the actual VO2max and HRmax are used.

For the third aim, 48 participants performed various activities ranging from sedentary pursuits to vigorous exercise. The activities were split into three routines of six activities and each participant performed one routine. During each routine an Actigraph (right hip), Actical (left hip), Actiheart (chest), and AMP-331 (right ankle) were worn. Simultaneously, EE was measured by IC. The Actiheart HR algorithm was not significantly different from measured EE for any of the 18 activities (P ≥ 0.05). The Actiheart combined HR and activity algorithm was only significantly different from measured EE for vacuuming and ascending/descending stairs (P < 0.05). All remaining prediction equations, for the devices examined, over- or underestimated EE for at least seven activities. The Actiheart HR algorithm provided the best estimate of EE over a wide range of activities. The Actical and Actigraph tended to overestimate walking and sedentary activities and underestimate most other activities.

For the fourth aim, 48 participants performed various activities (sedentary, light, moderate, and vigorous intensities) that were split into three routines of six activities. Each participant performed one routine. During each test the participants wore an Actigraph accelerometer and EE was measured by IC. Forty-five tests were randomly selected for the development of the new equation, and 15 tests were used to cross-validate the new equation and compare against existing equations. For each activity the coefficient of variation (CV) of the counts per 10 seconds was calculated to determine if the activity was walking/running, or some other activity. If the CV ≤ 10 then a walking/running regression equation (relating counts·min-1 to METs) was used, while if the CV > 10 a lifestyle/leisure time physical activity (LTPA) regression was used. The new 2-regression model explained 73% of the variance in EE for walking/running, and 83.8% of the variance in EE for lifestyle/LTPA and it was within ± 0.84 METs of measured METs for each of the 17 activities performed (P ≥ 0.05). The new 2-regression model is a more accurate prediction of EE then the currently published regression equations using the Actigraph accelerometer.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Included in

Education Commons