本論文旨在使用加速度感測模組與心電感測器實現一代謝當量估測模式，並搭配殘差估測演算法，提升代謝當量估測之精準度。本論文演算法使用佩戴於手腕、腳踝及腰部之三軸加速度感測模組與心電感測器收集之訊號，估測使用者活動時的代謝當量(metabolic equivalents, METs)。本研究首先依照加速度訊號特徵建構活動分類器，用於辨別生活中常見五種動作類型（全身不動、手腰腳活動、手腳活動、手部運動、腳部運動）。在各種活動類別中，基於加速度訊號及心電訊號之特徵，利用依序向前搜尋法(sequential forward selection, SFS)分別建立線性迴歸模型與類神經網路模型以估測代謝當量或是估測值與實際代謝當量的殘差，利用這兩種模型組合成為六種估測模式：(1)線性代謝當量估測模型、(2)類神經網路估測模型、(3)以類神經網路估測殘差搭配線性代謝當量估測模型、(4)以線性殘差估測模型搭配類神經網路代謝當量估測模型、(5)以線性殘差估測模型搭配線性代謝當量估測模型、(6)以類神經網路殘差估測模型搭配類神經網路代謝當量估測模型。本論文以判定係數(R2)以及代謝當量平均估測誤差來驗證所提出之方法之有效性，使用類神經網路代謝當量估測模型搭配類神經網路殘差估測模型進行代謝當量估測之R2=0.9565，相較於使用類神經網路代謝當量估測模型估測之R2=0.952與線性代謝當量估測模型之R2=0.9363相關性要來得高；而類神經網路代謝當量估測模型搭配類神經網路殘差估測模型之平均估測誤差為0.43±0.44 (METs)，相較類神經網路代謝當量估測模型之平均誤差0.44±0.46 (METs)，與線性代謝當量估測模型之平均估測誤差0.52±0.51 (METs)，其準確度更高。

This thesis presents a metabolic equivalents (METs) model with a residual estimation model for METs estimation. The proposed approach classifies most common daily activities into five types of categories. Features derived from electrocardiogram (ECG) and acceleration signals are first selected by sequential forward selection (SFS) strategy. The selected features are then used to construct a linear METs regression model or a neural network for preliminary METs estimation. These selected features are also constructed to improve the estimation accuracy by using the residual error, the difference between the estimated result of the METs estimation model and the actual METs. The final estimated METs are obtained by combining the outputs of the METs estimation model and the residual error compensation obtained by the residual estimation model. Using the two models to construct six combinations: Model1: linear regression model, Model 2: an NN METs estimation model, Model 3: a combination of a linear METs regression model and an NN METs residual error estimation model, Model 4: a combination of an NN METs estimation model and a linear residual error estimation model, Model 5: a combination of a linear METs regression model and a linear residual error estimation model, Model 6: a combination of an NN METs estimation model and an NN METs residual error estimation model. In order to validate the proposed approach, we compared the coefficients of determination (R2) and the mean absolute errors of the six models. In our experiments, the R2 of Model 6 is R2=0.9565 which is higher than Model 2 (R2=0.952). This result is also higher than Model 1 (R2=0.9363). The MAE of Model 6 is 0.43±0.44 METs, which is higher than using Model 2. This result is also better than Model 1, 0.52±0.51 METs.

參考文獻
O. H. Franco, C. de Laet, A. Peeters, J. Jonker, J. Mackenbach, and W. Nusselder, “Effects of physical activity on life expectancy with cardiovascular disease,” Arch Intern Med, vol. 165, no. 20, pp. 2355-2360, 2005.
I. M. Lee, H. D. Sesso, and R. S. Paffenbarger, “Physical activity and coronary heart disease risk in men does the duration of exercise episodes predict risk,” Circulation, vol. 102, no. 9, pp. 981-986, 2000.
B. A. Calton, R. Z. Stolzenberg-Solomon, S. C. Moore, A. Schatzkin, C. Schairer, D. Albanes, and M. F. Leitzmann, “A prospective study of physical activity and the risk of pancreatic cancer among women (United States),” BioMed Central Cancer, vol. 8, no. 1, pp. 63-72, 2008.
W. H. Pan, H. J. Wu, C. J. Yeh, S. Y. Chuang, H.Y. Chang, N. H. Yeh, and Y. T. Hsieh, “Diet and health trends in Taiwan: comparison of two nutrition and health surveys from 1993-1996 and 2005- 2008,” Asia Pac J Clin Nutr, vol. 20, no. 2, pp. 238-250, 2011
M. P. Rothney, E. V. Schaefer, M. M. Neumann, L. Choi, and K. Y. Chen, “Validity of physical activity intensity predictions by Actigraph, Actical, and RT3 accelerometers,” Obesity Society, vol. 16, no. 8, pp. 1946-1952, 2008.
S. E. Crouter, J. R. Churilla, and D. R. Bassett, “Estimating energy expenditure using accelerometers,” Eur. Journal of Applied Physiology, vol. 98, no. 6, pp. 601-612, 2006.
S. M. Slootmaker, C. A Paw, A. J. Schuit, W. Van. Mechelen, and L. L. J. Koppes, “Concurrent validity of the PAM accelerometer relative to the MTI Actigraph using oxygen consumption as a reference,” Medicine & Science in Sports & Exercise, vol. 19, no. 1, pp. 36-43, 2009.
K. Y. Chen and D. R. Bassett, “The technology of accelerometry-based activity monitors: current and future,” Medicine and Science in Sports and Exercise, vol. 37, no. 11, pp. S490-S500, 2005.
H. J. Montoye, R. Washburn, S. Servais, A. Ertl, J. G. Webster, and F. J. Nagle, “Estimation of energy expenditure by a portable accelerometer,” Medicine and Science in Sports and Exercise, vol. 15, pp. 403-407, 1982.
A. M. Swartz, S. J. Strath, D. R. Bassett, W. L. O’brien, G. A. King, and B. E. Ainsworth, “Estimation of energy expenditure using CSA accelerometers at hip and wrist sites,” Medicine and Science in Sports and Exercise, vol. 32, no. 9, pp. S450-S456, 2000.
P. S. Freedson, E. Melanson, and J. Sirard, “Calibration of the computer science and applications, Inc. accelerometer,” Medicine and Science in Sports and Exercise, vol. 30, no. 5, pp. 777-781, 1998.
D. Hendelman, K. C. Baggett, E. Debold, and P. Freedson, “Validity of accelerometry for the assessment of moderate intensity physical activity in the field,” Medicine and Science in Sports and Exercise, vol. 32, no. 9, pp. S442-S449, 2000.
M. P. Rotheny, M. Neuman, A. Beziat, and K. Y. Chen, “Validity of physical activity intensity predictions by Actigraph, Actical, and RT3 accelerometers,” Obesity, vol. 16, no. 8, pp. 1946-1952, 2008.
M. J. Karvonen, E. Kentala, and O. Mustala, “The effects of training on heart rate: A longitudinal study,” Ann Med Exp Biol Fenn, vol. 35, pp. 307-315, 1957.
D. P. Swain and B. C. Leutholtz, “Heart rate reserve is equivalent to %VO2 reserve, not to %VO2max,” Medicine and Science in Sports and Exercise, vol. 29, no. 3, pp. 410-414, 1997.
D. P. Swain, B. C. Leutholtz, M. E. King, L. A. Haas, and J. D. Branch, “Relationship between % heart rate reserve and % VO2 reserve in treadmill exercise,” Medicine and Science in Sports and Exercise, vol. 30, pp. 318-321, 1998.
N. M. Moyna, R. J. Robertson, C. L. Meckes, J. A. Peoples, N. B. Millich, and P. D. Thompson, “Intermodal comparison of energy expenditure at exercise intensities corresponding to the perceptual preference range,” Medicine and Science in Sports and Exercise, vol. 33, pp. 1404-1410, 2001.
G. P. Zhang, “Time series forecasting using a hybrid ARIMA and neural network model,” Neurocomputing, vol. 50, pp. 159-175, 2003.
S. M. Al-Alawi, S. A. Abdul-Wahab, and C. S. Bakheit, “Combining principal component regression and artificial neural networks for more accurate predictions of ground-level ozone,” Environmental Modelling & Software, vol. 23, no. 4, pp. 396-403, 2008.
S. J. Preece, J. Y. Goulermas, L. P. J. Kenney, D. Howard, K. Meijer, and R. Crompton, “Activity identification using body-mounted sensors-A review of classification techniques” Physiological Measurement, vol. 30, no. 4, pp. R1-R33, 2009.
E. M. Munguia, “Using Machine Learning for Real-time Activity Recognition and Estimation of Energy Expenditure”, Diss. Massachusetts Institute of Technology, 2008.
A. Whitney, “A direct method of nonparametric measurement selection,” IEEE Transactions on Computers, vol. 20, pp.1100-1103, 1971.
L. Breiman, J. Friedman, C. J. Stone, and R. A. Olshen, “Classification and regression trees,” CRC press, 1984.
E. K. Chong and S. H. Zak, “An introduction to optimization,” John Wiley & Sons, 2013.
K. Hornik, M. Stinchcombe, and H. White, “Multiplayer feedforward networks are universal approximators,” Neural Networks, vol.2, no. 5, pp.359-366, 1989.
J. J. More, “The Levenberg-Marquardt algorithm: implementation and theory,” Numerical analysis. Springer Berlin Heidelberg, pp. 105-116, 1978.
戴久永，統計概念與方法，民國88年8月。
M. Luštrek, B. Cvetković, and S. Kozina, “Energy expenditure estimation with wearable accelerometers,” in Proceedings of IEEE International Symposium on Circuits and Systems, pp. 5-8, 2012.
B. Cvetkovic, R. Milic, and M. Lustrek. “Estimating energy expenditure with multiple models using different wearable sensors,” Accepted by IEEE Journal of Biomedical and Health Informatics.
A. K. Bourke, F. Masse, A. Arami, K. Aminian, M. Healy, J. Nelson, and S. Coote, “Energy expenditure estimation using accelerometry and heart rate for multiple sclerosis and healthy older adults,” in Proceedings of 11th International Conference on Wearable and Implantable Body Sensor Networks Workshops, pp. 1-5, 2014.
J. W. Denton, “How good are neural networks for causal forecasting?,” The Journal of Business Forecasting, vol. 14, no. 2, pp. 17-20, 1995
P. F. Pai and C. S. Lin. “A hybrid ARIMA and support vector machines model in stock price forecasting,” Omega, vol. 33, no. 6, pp. 497-505, 2005.
C. V. Bouten, K. T. Koekkoek, M. Verduin, R. Kodde, and J. D. Janssen, “A triaxial accelerometer and portable data processing unit for assessment the assessment of daily physical activity,” IEEE Transactions on Biomedical Engineering, vol. 44, no. 3, pp. 136-147, 1997.
F. C. Chuang, “Development of inertial-sensor-based hand gesture recognition and physical activity energy expenditure estimation technologies,” Ph. D dissertation, National Cheng Kung University, Taiwan, 2012.