心臟疾病的發生多半與平日生活型態有關，根據先前的研究表明[1]，現代人
因工作多不動而造成罹患心臟疾病的風險增加 1.18 倍，而為了能夠檢測心臟疾病與平
時活動狀況，本文將建立一個能夠快速檢測的平台，分別辨識 AF 有無復發以及辨識人
體活動型態，用於了解心房顫動與活動型態的關係。
為了觀察心房顫動復發的問題，且透過電燒後能減少復發的風險[2]，並能利用非
侵入式訊號觀察取得神經訊號，之後透過對交感神經訊號資料分析，能夠找出心 AF 動
復發與沒有復發之間統計上差別，能既方便又安全的方式診斷心 AF 動的問題。
心房顫動(atrial fibrillation; AF)是指心房快速且不規律的收縮，當心房無法
有效地收縮，造成血液流動不佳，增加血管產生血栓的風險。一旦血栓順著血流流到
腦部血管使腦部血管阻塞，則會造成腦中風。心 AF 動約會增加 5 倍腦中風的風險，而
心房顫動導致的中風，預後很差且復發率高。[3]
自主神經系統在調節心臟離子通道與心肌收縮扮演很重要的角色，而以往周邊交
感神經活動(SNA)在量測時，須透過侵入式的電極量測，在技術上有困難且容易產生動
作雜訊， 而透過標準 ECG 貼片電極，可檢測多種電生理信號。[4]
將提取的電生理訊號透過數位濾波與其他資料處理方法提取其訊號的特徵，並用
於訓練機器學習模型上或其他資料分析。
為了人體活動偵測分析，我們研發無線藍芽低功耗三軸加速度計穿戴裝置，透過
「物聯網」以手機 APP 為介面，測量人體活動加速度，再提取三軸加速度訊號的特
徵，透過支援向量機(support vector machine)的方式分析人體活動型態。
三軸加速度計(G-sensor)為記錄加速度變化資訊的微機電零件，在工程上廣泛應
用，再加上固態微機電系統(Micro Electro Mechanical Systems , MEMS)的發展，使
零件尺寸能越來越小，製作成本隨著技術與發展也越來越低，已被廣泛應用在穿戴式
裝置中收集人體活動資料，用以訓練人體活動偵測模型。
本文將藉由多項分析了解電燒後心房顫動復發與沒有復發在 SKNA 訊號上是否有統
ii
計上的差別;也透過低功耗無線藍芽三軸加速度計的資料訓練人體活動偵測模型。

The occurrence of heart disease is mostly related to the daily life
style. According to previous studies [1], modern people’s risk of heart
disease increases by 1.18 times due to more inactive at work. In order to
be able to detect heart disease and daily activities, this study will build
a platform for rapid detection to understand the relationship between
atrial fibrillation and activity patterns.
To measure atrial fibrillation recurrence and reduce the risk of atrial
fibrillation recurrence after ablation, we extracted nerve signal with
noninvasive electrode, then analyzed sympathetic nerve activity signal to
find out the difference between atrial fibrillation recurrence and no
recurrence, it is convenience and security method to diagnose atrial
fibrillation recurrence.
Atrial fibrillation is an irregular and often rapid heart rate, When
the atria cannot be effectively contracted, poor blood flow is caused,
which increases the risk of blood clots. Once the thrombus flows along the
bloodstream to the blood vessels in the brain and blocks the blood vessels
in the brain, it will cause a stroke. Atrial fibrillation increases the
risk of stroke by 5 times. Stroke caused by atrial fibrillation has a poor
prognosis and a high recurrence rate. [3]
The autonomic nervous system is important to modulate cardiac ion
channel and myocardial contractility. Sympathetic nerve activity can be
measured with invasive microneurography techniques, these are technically
difficult and easy to produce motion artifact, but with standard ECG patch
electrode, we can detect multiple electrophysiological signals. [4]
iv
The electrophysiological signal is extracted through digital filtering
and other data processing methods to extract the characteristics of the
signal, and used for training machine learning models or other data
analysis.
In order to recognize human activity, we designed a wireless Bluetooth
Low Energy three-axis accelerometer wearable device, through the "Internet
of Things" using the mobile phone APP as the interface to measure the
acceleration of human activities, and then through the support vector
machine to analyze the types of human activity.
Three-axis accelerometer (G-sensor) is a micro-electromechanical part
that records acceleration change information. It is widely used in
engineering. With the development of Micro Electromechanical Systems
(MEMS), the size of components can be smaller, and the production cost is
getting lower and lower with technology and development. It has been widely
used in wearable devices to collect human activity data to train human
activity recognition models.
In the study, we will find out the statistically difference between AF
recurrence and AF no recurrence after ablation, and we will also train
human activity recognition model with the data of Bluetooth low energy
wireless three-axis accelerometer device.

[1] M. Kivimäki et al., "Long working hours as a risk factor for atrial fibrillation: a multicohort study," European Heart Journal, vol. 38, no. 34, pp. 2621-2628, 2017, doi:
10.1093/eurheartj/ehx324.
[2] A. Verma et al., "Approaches to Catheter Ablation for Persistent Atrial Fibrillation,"
New England Journal of Medicine, vol. 372, no. 19, pp. 1812-1822, 2015, doi:
10.1056/NEJMoa1408288.
[3] H. W. Choi, J. A. Navia, and G. S. Kassab, "Stroke Propensity Is Increased under Atrial
Fibrillation Hemodynamics: A Simulation Study," PLOS ONE, vol. 8, no. 9, p. e73485,
2013, doi: 10.1371/journal.pone.0073485.
[4] T. Kusayama et al., "Simultaneous noninvasive recording of electrocardiogram and
skin sympathetic nerve activity (neuECG)," Nature Protocols, vol. 15, no. 5, pp. 1853-
1877, 2020/05/01 2020, doi: 10.1038/s41596-020-0316-6.
[5] C. E. Chiang et al., "2016 Guidelines of the Taiwan Heart Rhythm Society and the
Taiwan Society of Cardiology for the management of atrial fibrillation," (in eng), J
Formos Med Assoc, vol. 115, no. 11, pp. 893-952, Nov 2016, doi:
10.1016/j.jfma.2016.10.005.
[6] M. G. Miglis, "Chapter 12 - Sleep and the Autonomic Nervous System," in Sleep and
Neurologic Disease, M. G. Miglis Ed. San Diego: Academic Press, 2017, pp. 227-244.
[7] A. Doytchinova et al., "Simultaneous noninvasive recording of skin sympathetic nerve
activity and electrocardiogram," Heart Rhythm, vol. 14, no. 1, pp. 25-33, 2017/01/01/
2017, doi: https://doi.org/10.1016/j.hrthm.2016.09.019.
[8] P. Castiglioni, A. Faini, G. Parati, and M. D. Rienzo, "Wearable Seismocardiography," in
2007 29th Annual International Conference of the IEEE Engineering in Medicine and
Biology Society, 22-26 Aug. 2007 2007, pp. 3954-3957, doi:
10.1109/IEMBS.2007.4353199.
[9] N. Mora, F. Cocconcelli, G. Matrella, and P. Ciampolini, "Accurate Heartbeat Detection
on Ballistocardiogram Accelerometric Traces," IEEE Transactions on Instrumentation
and Measurement, vol. 69, no. 11, pp. 9000-9009, 2020, doi:
10.1109/TIM.2020.2998644.
[10] E. García-Gonzalo, Z. Fernández-Muñiz, P. J. Garcia Nieto, A. Sánchez, and M.
Menéndez, "Hard-Rock Stability Analysis for Span Design in Entry-Type Excavations
with Learning Classifiers," Materials, vol. 9, p. 531, 06/29 2016, doi:
10.3390/ma9070531.
[11] D. Anguita, A. Ghio, L. Oneto, X. Parra, and J. L. Reyes-Ortiz, "Human Activity
Recognition on Smartphones Using a Multiclass Hardware-Friendly Support Vector
Machine," in Ambient Assisted Living and Home Care, Berlin, Heidelberg, J. Bravo, R.
63
Hervás, and M. Rodríguez, Eds., 2012// 2012: Springer Berlin Heidelberg, pp. 216-223.
[12] X. Gao, H. Luo, Q. Wang, F. Zhao, L. Ye, and Y. Zhang, "A Human Activity Recognition
Algorithm Based on Stacking Denoising Autoencoder and LightGBM," Sensors, vol. 19,
no. 4, p. 947, 2019. [Online]. Available: https://www.mdpi.com/1424-8220/19/4/947.
[13] J. L. Reyes-Ortiz, A. Ghio, X. Parra, D. Anguita, J. Cabestany, and A. Català, "Human
Activity and Motion Disorder Recognition: towards smarter Interactive Cognitive
Environments," in ESANN, 2013.
[14] T. Kusayama et al., "Skin sympathetic nerve activity and the temporal clustering of
cardiac arrhythmias," (in eng), JCI Insight, vol. 4, no. 4, Feb 21 2019, doi:
10.1172/jci.insight.125853.
[15] S.-H. Shin et al., "Left Atrial Volume Is a Predictor of Atrial Fibrillation Recurrence
After Catheter Ablation," Journal of the American Society of Echocardiography, vol.
21, no. 6, pp. 697-702, 2008/06/01/ 2008, doi:
https://doi.org/10.1016/j.echo.2007.10.022.
[16] STUDENT, "THE PROBABLE ERROR OF A MEAN," Biometrika, vol. 6, no. 1, pp. 1-25,
1908, doi: 10.1093/biomet/6.1.1.
[17] "nordicsemi.com." https://www.nordicsemi.com/products/nrf51822 (accessed 04
October, 2021).
[18] L. Novak, P. Neuzil, J. Li, and M. Woo, "Ultrasensitive MEMS-based Inertial System,"
pp. 552-554, 10/01 2009, doi: 10.1109/ICSENS.2009.5398301.
[19] "I2C-bus specification and user manual Rev. 6."
https://web.archive.org/web/20160804230952/http://www.nxp.com:80/documents/
user_manual/UM10204.pdf (accessed 4 April, 2014).
[20] C. Kim et al., 11.4 A 512Gb 3b/cell 64-stacked WL 3D V-NAND flash memory. 2017, pp.
202-203.
[21] " SPI Block Guide v3.06; Motorola/Freescale/NXP; 2003."
https://web.archive.org/web/20150413003534/http://www.ee.nmt.edu/~teare/ee3
08l/datasheets/S12SPIV3.pdf (accessed 04 February, 2004).
[22] M. Honkanen, A. Lappetelainen, and K. Kivekas, "Low end extension for Bluetooth,"
in Proceedings. 2004 IEEE Radio and Wireless Conference (IEEE Cat. No.04TH8746),
22-22 Sept. 2004 2004, pp. 199-202, doi: 10.1109/RAWCON.2004.1389107.
[23] "developer.android.com."
https://developer.android.com/guide/components/activities/activity-lifecycle
(accessed 04 October, 2021).
[24] D. Anguita, A. Ghio, L. Oneto, X. Parra, and J. L. Reyes-Ortiz, "A Public Domain Dataset
for Human Activity Recognition using Smartphones," in ESANN, 2013.
[25] J.-L. Reyes-Ortiz, L. Oneto, A. Samà, X. Parra, and D. Anguita, "Transition-Aware
Human Activity Recognition Using Smartphones," Neurocomputing, vol. 171, pp. 754-
64
767, 2016/01/01/ 2016, doi: https://doi.org/10.1016/j.neucom.2015.07.085.
[26] T. Watanabe. "github." https://github.com/takumiw/Deep-Learning-for-HumanActivity-Recognition (accessed 27 January, 2021)