背景：生理指標預測的功能在現今坊間不少穿戴式裝置中存在，而其運算通常是以統計或是常模式的推估。本研究將嘗試以機器學習建置一個在預測心跳率上更為準確的演算機制，並比較機器學習與統計模型的差異。方法：本研究納入了1位受試者從2019年7月2日至2021年3月28日的全日生活數據（有效資料374筆），並以此建構一個機器學習和一個統計模型。機器學習模型會以KNIME數據處理平台中的隨機森林演算法建置，並以留一驗證法(leave-one-out cross-validation, LOOCV)進行交叉驗證。為檢驗適合建置模型的訓練集樣本數，一開始會先以3個樣本投入訓練集，然後逐一增加，直到所有樣本都投入訓練與驗證，。統計模型會以SPSS 17.0中的逐步回歸法建置。結果：越多的樣本數投入訓練的機器學習模型解釋力更高（判定係數更高）、誤差更小（絕對誤差值較低）、表現更穩定（標準誤較低）。而在訓練的樣本數一樣的條件下，機器學習模型比統計模型有更好的表現。機器學習模型的相關係數和判定係數（r= 0.65、R2= 0.42）高於統計模型(r= 0.55、R2= 0.29)，機器學習模型的標準誤(SE= 1.28)也低於統計模型(SE= 1.40)。結論：透過選擇適當的訓練樣本類別以及足夠的資料量，機器學習能夠比傳統統計方法有更好的預測表現。在本研究中，機器學習在預測心跳率的表現上比統計模型還要更準確。

Prediction of physiology status in commercial wearable devices often performed with statistical analysis (SA) or norm-referenced evaluation. To make prediction more reliable, adopt a fitter algorithm is necessary. Therefore, the aim of current study attempts to obtain a machine learning (ML) model for heart rate (HR) prediction and assess the difference between ML and SA model. Present research involved one participant’s daily living data (from July 2nd, 2019 to March 28th, 2021) (valid sample n = 374) in establishing a ML model and a SA model. ML model is built with random forest on KNIME platform, and it would be trained with sample size n = 3 to n = 374 to evaluate the performance via different size of training set. SA model is built with stepwise regression on SPSS 17.0. The r, R2 and standard error (SE) of two models would be compared. Results showed that the ML model with larger training set performed better interpretability, lower error and higher stability. Besides, the ML model had better performance than the SA model when established with same sample size. To conclude, this study may be useful in explaining the importance of features selection and sample size in ML, as well as in providing heart rate prediction with a better algorithm instead of a SA model.

Achten, J., & Jeukendrup, A. E. (2003). Heart rate monitoring. Sports medicine, 33(7), 517-538.
Borresen, J., & Lambert, M. I. (2009). The quantification of training load, the training response and the effect on performance. Sports medicine, 39(9), 779-795.
Bruce, R., Kusumi, F., Culver, B., & Butler, J. (1975). Cardiac limitation to maximal oxygen transport and changes in components after jogging across the US. Journal of applied physiology, 39(6), 958-964.
Buchheit, M. (2014). Monitoring training status with HR measures: do all roads lead to Rome? Frontiers in physiology, 5, 73.
Budgett, R. (1998). Fatigue and underperformance in athletes: the overtraining syndrome. British journal of sports medicine, 32(2), 107-110.
Cincin, A., Sari, I., Oğuz, M., Sert, S., Bozbay, M., Ataş, H., . . . Basaran, Y. (2015). Effect of acute sleep deprivation on heart rate recovery in healthy young adults. Sleep and Breathing, 19(2), 631-636.
Dehghani, M., Abubakar, A. M., & Pashna, M. (2020). Market-driven management of start-ups: The case of wearable technology. Applied computing and informatics.
Derman, W., Schwellnus, M., Lambert, M., Emms, M., Sinclair-Smith, C., Kirby, P., & Noakes, T. (1997). The ‘worn-out athlete’: a clinical approach to chronic fatigue in athletes. Journal of sports sciences, 15(3), 341-351.
Education, M. o. (2018). 107年運動現況調查案結案報告書. Taipei, Taiwan
Edwards, S. (1994). The heart rate monitor book. In: LWW.
Esculier, J.-F., Bouyer, L. J., Dubois, B., Fremont, P., Moore, L., McFadyen, B., & Roy, J.-S. (2018). Is combining gait retraining or an exercise programme with education better than education alone in treating runners with patellofemoral pain? A randomised clinical trial. British journal of sports medicine, 52(10), 659-666.
Foster, C., Florhaug, J. A., Franklin, J., Gottschall, L., Hrovatin, L. A., Parker, S., . . . Dodge, C. (2001). A new approach to monitoring exercise training. The Journal of Strength & Conditioning Research, 15(1), 109-115.
Foster, C., Rodriguez-Marroyo, J. A., & De Koning, J. J. (2017). Monitoring training loads: the past, the present, and the future. International journal of sports physiology and performance, 12(s2), S2-2-S2-8.
Gough, C. (2018). Running & Jogging-Statistics & Facts. In: Statista.
Halson, S. L. (2014). Monitoring training load to understand fatigue in athletes. Sports medicine, 44(2), 139-147.
Hynynen, E., Vesterinen, V., Rusko, H., & Nummela, A. (2010). Effects of moderate and heavy endurance exercise on nocturnal HRV. International journal of sports medicine, 31(06), 428-432.
Irwin, M., Smith, T. L., & Gillin, J. C. (1992). Electroencephalographic sleep and natural killer activity in depressed patients and control subjects. Psychosomatic Medicine.
Jeukendrup, A., Hesselink, M., Snyder, A., Kuipers, H., & Keizer, H. (1992). Physiological changes in male competitive cyclists after two weeks of intensified training. International journal of sports medicine, 13(07), 534-541.
Lambert, M., Mbambo, Z., & Gibson, A. S. C. (1998). Heart rate during training and competition for longdistance running. Journal of sports sciences, 16(sup1), 85-90.
Leger, D., Metlaine, A., & Choudat, D. (2005). Insomnia and sleep disruption: relevance for athletic performance. Clinics in sports medicine, 24(2), 269-285.
LUCÍA, A., HOYOS, J., SANTALLA, A., EARNEST, C., & CHICHARRO, J. L. (2003). Tour de France versus Vuelta a Espana: which is harder? Medicine & Science in Sports & Exercise, 35(5), 872-878.
Lynch, S. L., & Hoch, A. Z. (2010). The female runner: gender specifics. Clinics in sports medicine, 29(3), 477-498.
Martin, B. J. (1981). Effect of sleep deprivation on tolerance of prolonged exercise. European journal of applied physiology and occupational physiology, 47(4), 345-354.
Meeusen, R., Watson, P., Hasegawa, H., Roelands, B., & Piacentini, M. F. (2006). Central fatigue. Sports medicine, 36(10), 881-909.
Ometov, A., Shubina, V., Klus, L., Skibińska, J., Saafi, S., Pascacio, P., . . . Chukhno, O. (2021). A survey on wearable technology: History, state-of-the-art and current challenges. Computer Networks, 193, 108074.
Pescatello, L. S., Riebe, D., & Thompson, P. D. (2014). ACSM's guidelines for exercise testing and prescription: Lippincott Williams & Wilkins.
Pichot, V., Roche, F., Gaspoz, J.-M., Enjolras, F., Antoniadis, A., Minini, P., . . . Barthelemy, J. C. (2000). Relation between heart rate variability and training load in middle-distance runners. Medicine and science in sports and exercise, 32(10), 1729-1736.
Roper, J. L., Harding, E. M., Doerfler, D., Dexter, J. G., Kravitz, L., Dufek, J. S., & Mermier, C. M. (2016). The effects of gait retraining in runners with patellofemoral pain: a randomized trial. Clinical biomechanics, 35, 14-22.
Shipway, R., & Holloway, I. (2010). Running free: Embracing a healthy lifestyle through distance running. Perspectives in public health, 130(6), 270-276.
Stagno, K. M., Thatcher, R., & Van Someren, K. A. (2007). A modified TRIMP to quantify the in-season training load of team sport players. Journal of sports sciences, 25(6), 629-634.
Stewart, M. (2019). The actual difference between statistics and machine learning. Towards Data Science. Retrieved from.
Stone, J. D., Rentz, L. E., Forsey, J., Ramadan, J., Markwald, R. R., & Finomore, V. S. (2020). Evaluations of commercial sleep technologies for objective monitoring during routine sleeping conditions. Nature and Science of Sleep, 12, 821.
Thompson, W. R. (2015). Worldwide survey of fitness trends for 2018: the CREP edition. ACSM's Health & Fitness Journal, 19(6), 9-18.
Thompson, W. R. (2016). Worldwide survey of fitness trends for 2017. ACSMs Health Fit J, 20(6), 8-17.
Thompson, W. R. (2017). Worldwide survey of fitness trends for 2018: the CREP edition. ACSM's Health & Fitness Journal, 21(6), 10-19.
Thompson, W. R. (2018). Worldwide survey of fitness trends for 2019. ACSM's Health & Fitness Journal, 22(6), 10-17.
Thompson, W. R. (2019). Worldwide survey of fitness trends for 2020. ACSM's Health & Fitness Journal, 23(6), 10-18.
Thompson, W. R. (2021). Worldwide Survey of Fitness Trends for 2021. ACSM's Health & Fitness Journal, 25(1), 10-19. Retrieved from https://journals.lww.com/acsm-healthfitness/Fulltext/2021/01000/Worldwide_Survey_of_Fitness_Trends_for_2021.6.aspx. doi:10.1249/fit.0000000000000631
Wallace, L., Slattery, K., & Coutts, A. J. (2014). A comparison of methods for quantifying training load: relationships between modelled and actual training responses. European journal of applied physiology, 114(1), 11-20.
Xiao, F., Chen, Y., Yuchi, M., Ding, M., & Jo, J. (2010). Heart rate prediction model based on physical activities using evolutionary neural network. Paper presented at the 2010 Fourth International Conference on Genetic and Evolutionary Computing.
Ying, X. (2019). An overview of overfitting and its solutions. Paper presented at the Journal of Physics: Conference Series.