성능 개선을 기대해 볼 수 있다. 또한, 본 연구에서는 데이터 증강 기법이 성능 향상에 기여하는 지 확인하기 위해 각각의 방법들을 따로 따로 사용했다. 만약 성능 향상이 확인된 데이터 증강 기법들을 동시에 사용하여 모델을 학습한다면 성능이 향상될 수 있다.
참고 문헌
[1] Zewen Li, Fan Liu, Wenjie Yang, Shouheng Peng, and Jun Zhou. A survey of convolutional neural networks: analysis, applications, and prospects. IEEE transactions on neural networks and learning systems, 2021.
[2] Serkan Kiranyaz, Onur Avci, Osama Abdeljaber, Turker Ince, Moncef Gabbouj, and Daniel J Inman. 1d convolutional neural networks and applications: A survey. Mechanical systems and signal processing, 151:107398, 2021.
[3] Sumbal Maqsood, Shuxiang Xu, Son Tran, Saurabh Garg, Matthew Springer, Mohan Karunanithi, and Rami Mohawesh. A survey: From shallow to deep machine learning approaches for blood pressure estimation using biosensors.
Expert Systems with Applications, page 116788, 2022.
[4] Gašper Slapničar, Nejc Mlakar, and Mitja Luštrek. Blood pressure estima- tion from photoplethysmogram using a spectro-temporal deep neural network.
Sensors, 19(15):3420, 2019.
[5] 통계청, 신현영, et al. 2018 년 사망원인통계. 대한의사협회지, 63(5):286–297, 2020.
[6] Gregory A Roth, Degu Abate, Kalkidan Hassen Abate, Solomon M Abay, Cris- tiana Abbafati, Nooshin Abbasi, Hedayat Abbastabar, Foad Abd-Allah, Jemal
mortality for 282 causes of death in 195 countries and territories, 1980–2017:
a systematic analysis for the global burden of disease study 2017. The Lancet, 392(10159):1736–1788, 2018.
[7] Satoshi Umemura, Hisatomi Arima, Shuji Arima, Kei Asayama, Yasuaki Dohi, Yoshitaka Hirooka, Takeshi Horio, Satoshi Hoshide, Shunya Ikeda, Toshihiko Ishimitsu, et al. The japanese society of hypertension guidelines for the man- agement of hypertension (jsh 2019). Hypertension Research, 42(9):1235–1481, 2019.
[8] Remo Lazazzera, Yassir Belhaj, and Guy Carrault. A new wearable device for blood pressure estimation using photoplethysmogram. Sensors, 19(11):2557, 2019.
[9] Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. nature, 521(7553):436–444, 2015.
[10] Mohammad Kachuee, Mohammad Mahdi Kiani, Hoda Mohammadzade, and Mahdi Shabany. Cuffless blood pressure estimation algorithms for continu- ous health-care monitoring. IEEE Transactions on Biomedical Engineering, 64(4):859–869, 2016.
[11] Sanghyun Baek, Jiyong Jang, and Sungroh Yoon. End-to-end blood pressure prediction via fully convolutional networks. IEEE Access, 7:185458–185468, 2019.
[12] Kirk Shelley, Stacey Shelley, and Carol Lake. Pulse oximeter waveform: pho- toelectric plethysmography. Clinical monitoring, 2, 2001.
[13] Jessica E Wagenseil and Robert P Mecham. Elastin in large artery stiffness and hypertension. Journal of cardiovascular translational research, 5(3):264–
273, 2012.
[14] Manuja Sharma, Karinne Barbosa, Victor Ho, Devon Griggs, Tadesse Ghir- mai, Sandeep K Krishnan, Tzung K Hsiai, Jung-Chih Chiao, and Hung Cao.
Cuff-less and continuous blood pressure monitoring: a methodological review.
Technologies, 5(2):21, 2017.
[15] Charalambos Vlachopoulos, Michael O’Rourke, and Wilmer W Nichols. Mc- Donald’s blood flow in arteries: theoretical, experimental and clinical principles.
CRC press, 2011.
[16] J Crighton Bramwell and Archibald Vivian Hill. The velocity of pulse wave in man. Proceedings of the Royal Society of London. Series B, Containing Papers of a Biological Character, 93(652):298–306, 1922.
[17] Muhamad Khairul Bin Ali Hassan, MY Mashor, NF Nasir, and S Mohamed.
Measuring blood pressure using a photoplethysmography approach. In 4th Kuala Lumpur International Conference on Biomedical Engineering 2008, pages 591–594. Springer, 2008.
[18] Mico Yee-Man Wong, Carmen Chung-Yan Poon, and Yuan-Ting Zhang. An evaluation of the cuffless blood pressure estimation based on pulse transit time technique: a half year study on normotensive subjects. Cardiovascular Engi- neering, 9(1):32–38, 2009.
[19] Shrimanti Ghosh, Ankur Banerjee, Nilanjan Ray, Peter W Wood, Pierre Boulanger, and Raj Padwal. Continuous blood pressure prediction from pulse transit time using ecg and ppg signals. In 2016 IEEE Healthcare Innova- tion Point-Of-Care Technologies Conference (HI-POCT), pages 188–191. IEEE, 2016.
[20] Yi-Yen Hsieh, Ching-Da Wu, Shey-Shi Lu, and Yu Tsao. A linear regression model with dynamic pulse transit time features for noninvasive blood pressure prediction. In 2016 IEEE Biomedical Circuits and Systems Conference (Bio- CAS), pages 604–607. IEEE, 2016.
[21] Bing Zhang, Zhiyao Wei, Jiadong Ren, Yongqiang Cheng, and Zhangqi Zheng.
An empirical study on predicting blood pressure using classification and regres- sion trees. IEEE access, 6:21758–21768, 2018.
[22] Paul C-P Chao, Chih-Cheng Wu, Duc Huy Nguyen, Ba-Sy Nguyen, Pin-Chia Huang, and Van-Hung Le. The machine learnings leading the cuffless ppg blood pressure sensors into the next stage. IEEE Sensors Journal, 21(11):12498–
12510, 2021.
[23] XF Teng and YT Zhang. Continuous and noninvasive estimation of arterial blood pressure using a photoplethysmographic approach. InProceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No. 03CH37439), volume 4, pages 3153–3156.
IEEE, 2003.
[24] Satomi Suzuki and Koji Oguri. Cuffless blood pressure estimation by error- correcting output coding method based on an aggregation of adaboost with a photoplethysmograph sensor. In 2009 Annual international conference of the IEEE engineering in medicine and biology society, pages 6765–6768. IEEE, 2009.
[25] Arata Suzuki and Kazuteru Ryu. Feature selection method for estimating sys- tolic blood pressure using the taguchi method.IEEE Transactions on Industrial Informatics, 10(2):1077–1085, 2013.
[26] Mohamad Kachuee, Mohammad Mahdi Kiani, Hoda Mohammadzade, and Mahdi Shabany. Cuff-less high-accuracy calibration-free blood pressure esti- mation using pulse transit time. In 2015 IEEE international symposium on circuits and systems (ISCAS), pages 1006–1009. IEEE, 2015.
[27] Daisuke Fujita, Arata Suzuki, and Kazuteru Ryu. Ppg-based systolic blood pressure estimation method using pls and level-crossing feature. Applied Sci- ences, 9(2):304, 2019.
[28] Francesco Lamonaca, Kurt Barbe, Yuriy Kurylyak, Domenico Grimaldi, Wendy Van Moer, Angelo Furfaro, and Vitaliano Spagnuolo. Application of the arti- ficial neural network for blood pressure evaluation with smartphones. In 2013 IEEE 7th International Conference on Intelligent Data Acquisition and Ad- vanced Computing Systems (IDAACS), volume 1, pages 408–412. IEEE, 2013.
[29] Shrimanti Ghosh, Ankur Banerjee, Nilanjan Ray, Peter W Wood, Pierre
prove blood pressure prediction from pulse transit time using recurrent neural network. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 935–939. IEEE, 2018.
[30] B Moody, G Moody, M Villarroel, G Clifford, and I Silva. Mimic-iii waveform database matched subset (version1. 0). physionet, 2020.
[31] Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, and Liang- Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In Pro- ceedings of the IEEE conference on computer vision and pattern recognition, pages 4510–4520, 2018.
[32] Jean-Bastien Grill, Florian Strub, Florent Altché, Corentin Tallec, Pierre Richemond, Elena Buchatskaya, Carl Doersch, Bernardo Avila Pires, Zhao- han Guo, Mohammad Gheshlaghi Azar, et al. Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural information processing systems, 33:21271–21284, 2020.
[33] Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimiza- tion. arXiv preprint arXiv:1412.6980, 2014.
Abstract
Calibrating PPG Based Blood Pressure Estimation Model Using Convolutional
Neural Networks
Haesung Kim Department of Industrial Engineering The Graduate School Seoul National University
Blood pressure estimation using PPG signals is a research field that estimates blood pressure from PPG signals collected non-invasively through optical methods. Unlike existing blood pressure estimation methods, it can estimate blood pressure contin- uously without user inconvenience in daily life. Recently, using deep learning tech- niques, studies to estimate blood pressure from the PPG signal itself in an end-to-end manner have been proposed. Although patient calibration, which allows different patients to be distinguished, has significantly improved blood pressure estimation performance, there are not many studies focusing on this.
In this paper, we propose a convolutional neural network model that uses patient calibration data as input. Calibration data consist of a PPG signal of 4 seconds and