Evaluation of custom energy expenditure models for SenseWear armband in manual wheelchair users

Abstract

Physical activity monitors are increasingly used to help the general population lead a healthy lifestyle by keeping track of their daily physical activity (PA) and energy expenditure (EE). However, none of the commercially available activity monitors can accurately estimate PA and EE in people who use wheelchairs as their primary means of mobility. Researchers have recently developed custom EE prediction models for manual wheelchair users (MWUs) with spinal cord injuries (SCIs) based on a commercial activity monitor--the SenseWear armband. This study evaluated the performance of two custom EE prediction models, including a general model and a set of activity-specific models among 45 MWUs with SCI. The estimated EE was obtained by using the two custom models and the default manufacturer's model, and it was compared with the gold standard measured by the K4b2 portable metabolic cart. The general, activity-specific, and default models had a mean signed percent error (mean +/- standard deviation) of -2.8 +/- 26.1%, -4.8 +/- 25.4%, and -39.6 +/- 37.8%, respectively. The intraclass correlation coefficient was 0.86 (95% confidence interval [CI] = 0.82 to 0.89) for the general model, 0.83 (95% CI = 0.79 to 0.87) for the activity-specific model, and 0.62 (95% CI = 0.16 to 0.81) for the default model. The custom models for the SenseWear armband significantly improved the EE estimation accuracy for MWUs with SCI.

Measurement of Physical Activity and Energy Expenditure in Wheelchair Users: Methods, Considerations and Future Directions

“…More detailed components and specifications of this activity monitor have been described elsewhere [ 36 ]. It is clear that the proprietary manufacturer’s algorithms intrinsic to the SWA device are not appropriate to predict EE in persons that use wheelchairs, with ICCs < 0.64 [ 40 , 70 , 71 ]. The overestimation of EE by the SWA manufacturer’s model is likely due to the movements typically performed by persons that use wheelchairs (e.g.…”

“…Researchers have developed new EE prediction models (SCI general and activity specific) for the SWA device that have been cross-validated [ 40 , 71 ]. Where MAE statistics are available [ 40 , 70 , 71 ] weighted means were calculated, with the SCI general (22.7%) and activity specific (18.2%) models performing significantly better than the manufacturer’s model (54.4%).…”

“…Researchers have developed new EE prediction models (SCI general and activity specific) for the SWA device that have been cross-validated [ 40 , 71 ]. Where MAE statistics are available [ 40 , 70 , 71 ] weighted means were calculated, with the SCI general (22.7%) and activity specific (18.2%) models performing significantly better than the manufacturer’s model (54.4%). Whilst these findings provide encouragement for the use of the SWA in persons that use wheelchairs with new prediction models, Conger et al [ 72 ] noticed that even when using the SCI general model, the SWA tended to overestimate EE (27 to 43%), whereas a wrist-mounted accelerometer more accurately predicted EE (9 to 25%) during wheelchair prolusion (Table 3 ).…”

“…Neither device is an appropriate tool for quantifying EE (<0.75) Hiremath et al . [ 70 ] c 45 (8 F) (C4–L4) IC EE (COSMED K4b 2 ) Estimated EE from SWA worn on the upper arm (manufacturer’s model) ICC = 0.64 (95% CI; 0.57–0.70) M ± E = 51.5 ± 31.6% MAE = 2.0 kcal · min -1 (59.2%) Estimated EE from SWA worn on the upper arm (SCI general model) ICC = 0.72 (95% CI; 0.66 – 0.77) M ± E = -10.4 ± 11.8% MAE = 0.9 kcal · min -1 (24.7%) Estimated EE from SWA worn on the upper arm (activity-specific model) ICC = 0.86 (95% CI; 0.82–0.88) M ± E = −9.6 ± 10.9% MAE = 0.6 kcal · min -1 (16.8%) Tsang et al [ 71 ] c 45 SCI b (6 F) (C5 – L5) IC EE (COSMED K4b 2 ) Participants performed 10 activities from a list that included a range of activities and exercise of differing intensities Estimated EE from SWA worn on the upper arm (manufacturer’smodel) ICC = 0.62 (95% CI; 0.16 – 0.81) M ± E = 39.6 ± 37.8% MAE = 43.3 ± 33.5% Estimated EE from SWA worn on the upper arm (SCI general model) ICC = 0.86 (95% CI; 0.82 – 0.89) M ± E = 2.8 ± 26.1% MAE = 20.6 ± 16.2% Estimated EE from SWA worn on the upper arm (activity-specific model) ICC = 0.83 (95% CI; 0.79 – 0.87) M ± E = 4.8 ± 25.4% MAE = 19.6 ± 16.8% Nightingale et al . [ 75 ] 15.…”

“…The SWA and PASIPD both performed particularly poorly in the prediction of PAEE, displaying considerable random error as demonstrated by the large 95% limits of agreement. It is noteworthy that the SWA used the aforementioned error-prone manufacturer’s model [ 40 , 70 , 71 ] but could be improved with the SCI general EE prediction model developed by Hiremath et al [ 70 ]. One limitation of the Tanhoffer et al [ 66 ] study is that the length of PA monitoring period for each prediction method varied compared to the criterion method.…”