隨著生活品質的提升，雖然使人們的平均壽命增加，卻也造成高齡化社會問題日益嚴重。帕金森氏症(Parkinson’s Disease, PD)為一種好發於中老年人的神經退化性疾病，目前尚無根治的方法，僅能透過藥物與手術減緩症狀，但所有藥物皆會有影響日常生活之副作用(如：幻覺、憂鬱症、睡眠障礙及低血壓等症狀)。
帕金森氏症之早期檢測仰賴Hoehn-Yahr分級表、帕金森氏症評估量表與核子醫學影像評估，前兩者為患者根據自身狀況所填寫之量表，導致缺乏一致性與公正性，後者會有輻射劑量過高與等待時間較長之問題，使醫師難以精準調配藥物劑量，影響治療效果。因此，如何研發一應用於帕金森氏症患者之步態檢測系統，使醫師在診斷時能有參考標準為一重要議題。
本研究提出一整合式的學習架構，其內容包含了深度學習、機器學習、資料選擇、特徵評估與資料平衡等機制。針對帕金森氏症患者的步態進行檢測，分析出正常人與帕金森式症患者的差異，期望能提供醫師一套在診斷時可參考的標準。本研究之資料來源為PhysioNet公開資料庫，資料為腳底壓力感測器之數據，在特徵萃取之部分，為了讓研究更符合目前穿戴式裝置之趨勢，本研究透過ANOVA找出一組(左右腳)最重要之感測器位置，並依此開發出22個特徵，包含了Force-domain features、Peak-domain features與Abnormality-domain features。為了減少數據中的雜訊，本研究開發了一套ANOVA with Recursive Reduction (ARR)機制，將會影響準確率之特徵移除，達到減少模型雜訊之目的。由於原始資料集的帕金森氏症患者與正常受試者存在數量上的差異，為了減少數據不平衡對訓練模型時的影響，本研究於架構中加入了特徵平衡演算法，以提升整體模型之可信度與穩定性。透過本研究之架構，整體模型之辨識率具明顯的提升，XGBoost模型之準確率可達97.32%，而CNN模型之準確率可達98.41%。

As a result of improvements in quality of life, the average human life expectancy has been increasing annually, leading to the aging problems by increasingly serious. Parkinson’s disease (PD) is a neurodegenerative disease that develops in middle-aged and older adults. No cure currently exists for the disease, and it can only be alleviated through surgery and use of medications. However, all of medications have side effects of pharmaceutical drugs that can affect people’s daily lives (e.g., symptoms including hallucinations, depression, sleep disorders, and low blood pressure).
Early diagnosis of PD relies on the Hoehn and Yahr scale, unified Parkinson's disease rating scale, and nuclear medicine imaging evaluation. The two aforementioned scales are completed by patients according to their own health conditions, which leads to a lack of consistency and objectivity for these methods. The imaging method requires an excessively high radiation dose and is associated with long waiting times, thus rendering dosage adjustment of medications difficult for doctors and thereby limiting its therapeutic effects. Therefore, the development of a gait detection method for patients with PD to serve as a reference for doctors when making diagnoses is a crucial research topic.
This study develops an integrated learning framework that includes deep learning, machine learning, data selection, feature evaluation, and data balancing mechanisms. This work performed gait detection on patients with PD and analyzed the difference in results between these patients and healthy individuals. The aim was to provide a data set that can serve as a reference for doctors when making a diagnosis. The data used in this study were foot pressure sensor data obtained from PhysioNet, which is an open database. In terms of feature extraction, to make research more congruent with current trends of wearable devices, this study conducted an ANOVA to identify sensor positions that exhibited the greatest difference (between left and right leg) as well as develop 22 features, including force-domain features, peak-domain features, and abnormality-domain features. To reduce noise in the data, this work developed a set of ANOVA with Recursive Reduction (ARR) mechanisms to remove features that might affect accuracy, thereby achieving the goal of reducing the noise for the model. The numbers of patients with PD and healthy individuals in the original database differed. To reduce the effect of an imbalance in data when training the models, this research incorporated a feature-balancing algorithm into the research framework, the overall validity and stability of the models are increased significantly. Additionally, the overall identification rate of the models exhibited a notable increase through application of the research framework developed in this study; specifically, the XGBoost and CNN models could achieve accuracy rates as high as 97.32% and 98.41%, respectively.

[1] 內政部統計處之人口三段年齡組成人數(2019/2)
(https://www.moi.gov.tw/stat/chart.aspx)
[2] M. M. Hoehn and M. D. Yahr, “Parkinsonism：Onset, Progression, and Mortality,” Neurology, Vol. 17, No. 5, pp. 427-442, 1967.
[3] N. Hirose, F. Xia, R. M. Martin, A. Sadeghian and S. Savarese, “Deep Visual MPC-Policy Learning for Navigation,” IEEE Robotics and Automation Letters (Early Access), 2019.
[4] M. Tan, J. Yu, H. Zhang, Y. Rui and D. Tao, “Image Recognition by Predicted User Click Feature with Multidomain Multitask Transfer Deep Network,” IEEE Transactions on Image Processing (Early Access), 2019.
[5] D. Ostrowski, “Artificial Intelligence with Big Data,” Proceedings of the First International Conference on Artificial Intelligence for Industries (AI4I), pp. 124-125, 2018.
[6] X. Wu, X. Chen, Y. Duan, S. Xu, N. Cheng and N. An, “A study on gait-based Parkinson's disease detection using a force sensitive platform,” Proceedings of the IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp. 2330-2332, 2017.
[7] Q. T. Ly, A. M. A. Handojoseno, M. Gilat, R. Chai, K. A. E. Martens, M. Georgiades, G. R. Naik, Y. Tran, S. J. G. Lewis and H. T. Nguyen, “Detection of gait initiation Failure in Parkinson's disease based on wavelet transform and Support Vector Machine,” Proceedings of the 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp. 3048-3051, 2017.
[8] J. S. Navya, S. Rajasree and B. Ullas, “Non-Invasive Fall Detection System for Parkinson's Disease,” Proceedings of the International CET Conference on Control, Communication, and Computing (IC4), pp. 157-160, 2017.
[9] T. D. Pham, “Texture Classification and Visualization of Time Series of Gait Dynamics in Patients With Neuro-Degenerative Diseases,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol. 26, No. 5, pp. 188-196, 2018.
[10] H. L. U. Thuc, P. V. Tuan and J. Hwang, “An effective video-based model for fall monitoring of the elderly,” Proceedings of the International Conference on System Science and Engineering (ICSSE), pp. 48-52, 2017.
[11] A. Elkholy, M. Hussein, W. Gomaa, D. Damen and E, Saba, “Efficient and Robust Skeleton-Based Quality Assessment and Abnormality Detection in Human Action Performance,” IEEE Journal of Biomedical and Health Informatics (Early Access), Vol. 14, No. 5, 2019.
[12] P. Tahafchi, R. Molina, J. A. Roper, K. Sowalsky, C. J. Hass, A. Gunduz, M. S. Okun and J. W. Judy, “Freezing-of-Gait detection using temporal, spatial, and physiological features with a support-vector-machine classifier,” 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp. 2867-2870, 2017.
[13] A. Baghdadi, “Application of Inertial Measurement Units for Advanced Safety Surveillance System Using Individualized Sensor Technology (ASSIST): A Data Fusion and Machine Learning Approach,” Proceedings of the IEEE International Conference on Healthcare Informatics (ICHI), pp. 450-451, 2018.
[14] N. Nazmi, M. A. A. Rahman, M. H. M. Ariff and S. A. Ahmad, “Generalization of ANN Model in Classifying Stance and Swing Phases of Gait using EMG Signals,” Proceedings of the IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES), pp. 461-466, 2018.
[15] Parkinson’s Foundation
(https://parkinson.org/)
[16] G. Dimauro, V. D. Nicola, V. Bevilacqua, D. Caivano and F. Girardi, “Assessment of Speech Intelligibility in Parkinson’s Disease Using a Speech-To-Text System,” IEEE Access, Vol. 5, pp. 22199-2220, 2017.
[17] M. Bravo, A. Bermeo, M. Huerta, C. Llumiguano, J. Bermeo, R. Clotet and A. Soto, “A system for finger tremor quantification in patients with Parkinson's disease,” Proceedings of the 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp. 3549-3552, 2017.
[18] J. Daneault, S. I. Lee, F. N. Golabchi, S. Patel, L. C. Shih, S. Paganoni and P. Bonato, “Estimating Bradykinesia in Parkinson's Disease with a Minimum Number of Wearable Sensors,” Proceedings of the IEEE/ACM International Conference on Connected Health: Applications, Systems and Engineering Technologies (CHASE), pp. 264-265, 2017.
[19] S. Tisch, J, C. Rothwell, P. Limousin, M. I. Hariz and D. M. Corcos, “The Physiological Effects of Pallidal Deep Brain Stimulation in Dystonia,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol. 15, No. 2, pp. 166-172, 2007.
[20] B. Muñoz, Y. J. Castaño-Pino, J. D. A. Paredes and A. Navarro, “Automated Gait Analysis using a Kinect Camera and Wavelets,” Proceedings of the IEEE 20th International Conference on e-Health Networking, Applications and Services (Healthcom), pp. 1-5, 2018.
[21] C. Lee, V. Fong, Y. Chu, L. Cheng, H. W. Chuang and C. Lo, “Automated Gait Analysis using a Kinect Camera and Wavelets,” International Conference on Machine Learning and Cybernetics (ICMLC), Vol. 2, pp. 462-466, 2018.
[22] D. D. Venuto, V. F. Annese, G. Mezzina and G. Defazio, “FPGA-Based Embedded Cyber-Physical Platform to Assess Gait and Postural Stability in Parkinson’s Disease,” IEEE Transactions on Components, Packaging and Manufacturing Technology, Vol. 8, No. 7, pp. 1167-1179, 2018.
[23] M. Vadovský and J. Paralič, “Parkinson's disease patients classification based on the speech signals” Proceedings of the IEEE 15th International Symposium on Applied Machine Intelligence and Informatics (SAMI), pp. 321-326 2017.
[24] J. Rusz, M. Novotný, J. Hlavnička, T. Tykalová and E. Růžička, “High-Accuracy Voice-Based Classification Between Patients With Parkinson’s Disease and Other Neurological Diseases May Be an Easy Task With Inappropriate Experimental Design,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol. 25, No. 8, pp. 1319-1321, 2017.
[25] C. G. Goetz, W. Poewe, O. Rascol, C. Sampaio, G. T. Stebbins, C. Counsell, N. Giladi, R. G. Holloway, C. G. Moore, G. K. Wenning, M. D. Yahr and L. Seidl, “Movement Disorder Society Task Force Report on the Hoehn and Yahr Staging Scale: Status and Recommendations,” Movement Disorder Society, Vol. 19, No. 9, pp. 1020-1028, 2004.
[26] C. Goetz, B. Tilley, S. Shaftman, G. Stebbins, S. Fahn, P. Martinez-Martin, W. Poewe, C. Sampaio, M. Stern, R. Dodel, B. Dubois, R. Holloway, J. Jankovic, J. Kulisevsky, A. Lang, A. Lees, S. Leurgans, P. LeWitt, D. Nyenhuis, C. Olanow, O. Rascol, A. Schrag, A. Teresi, J. Hilten and N. LaPelle, “Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results,” Movement Disorders, Vol. 23, No. 15, pp. 2129-2170, 2008.
[27] I. S. Klyuzhin, J. F. Fu, N, Shenkov, A. Rahmim and V, Sossi, “Use of Generative Disease Models for Analysis and Selection of Radiomic Features in PET,” IEEE Transactions on Radiation and Plasma Medical Sciences, Vol. 3, No. 2, pp. 178-191, 2019.
[28] V. Sossi, J. Cheng and I. S. Klyuzhin, “Imaging in Neurodegeneration: Movement Disorders,” IEEE Transactions on Radiation and Plasma Medical Sciences, Vol. 3, No. 3, pp. 262-274, 2019.
[29] M. Kim, J. H. Won, J. Youn and H. Park, “Joint-connectivity-based sparse canonical correlation analysis of imaging genetics for detecting biomarkers of Parkinson’s disease,” IEEE Transactions on Medical Imaging (Early Access), 2019.
[30] S. Aich, M. Sain, J. Park, K. Choi and H. Kim, “A text mining approach to identify the relationship between gait-Parkinson's disease (PD) from PD based research articles,” Proceedings of the International Conference on Inventive Computing and Informatics (ICICI), pp. 481-485, 2017.
[31] K. Wang, B. Li, D. Gu, K. Dai and L. Zhou, “A smartphone based system for freezing of gait monitoring for Parkinson's disease patients,” Proceedings of the IEEE 9th International Conference on Communication Software and Networks (ICCSN), pp. 1529-1533, 2017.
[32] K. D. Das, A. J. Saji and C. S. Kumar, “Frequency analysis of gait signals for detection of neurodegenerative diseases,” Proceedings of the 2017 International Conference on Circuit, Power and Computing Technologies (ICCPCT), pp. 1-6, 2017.
[33] E. Rovini, C. Maremmani and F. Cavallo, “How Wearable Sensors Can Support Parkinson's Disease Diagnosis and Treatment: A Systematic Review,” Frontiers in Neuroscience, Vol. 11, No. 11, pp. 1-41, 2017.
[34] T. T. Pham, S. T. Moore, S. J. G. Lewis, D. N. Nguyen, E. Dutkiewicz, A. J. Fuglevand, A. L. McEwan and P. H. Leong, “Freezing of Gait Detection in Parkinson's Disease: A Subject-Independent Detector Using Anomaly Scores,” IEEE Transactions on Biomedical Engineering, Vol. 64, No. 11, pp. 2719-2728, 2017.
[35] I. Mileti, M. Germanotta, E. D. Sipio, I. Imbimbo, A. Pacilli, C. Erra, M. Petracca, S. Rossi, Z. D. Prete, A. R. Bentivoglio, L. Padua and E. Palermo1, “Measuring Gait Quality in Parkinson’s Disease through Real-Time Gait Phase Recognition,” Sensors, Vol. 18, No. 3, pp. 1-16, 2018.
[36] M. Yoneyama, Y. Kurihara, K. Watanabe and H. Mitoma, “Accelerometry-Based Gait Analysis and Its Application to Parkinson's Disease Assessment— Part 2 : A New Measure for Quantifying Walking Behavior,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol. 21, No. 6, pp. 999-1005, 2013.
[37] T.T. Pham, S. T. Moore, S. J. G. Lewis, D. N. Nguyen, E. Dutkiewicz, A. J. Fuglevand, A. L. McEwan and P. H. W. Leong, “Freezing of Gait Detection in Parkinson's Disease: A Subject-Independent Detector Using Anomaly Scores,” IEEE Transactions on Biomedical Engineering, Vol. 64, No. 11, pp. 2719-2728, 2017.
[38] A. Procházka, M. Schätz, O. Ťupa, M. Yadollahi, O. Vysata and M. Walls, “The MS kinect image and depth sensors use for gait features detection,” Proceedings of the IEEE International Conference on Image Processing (ICIP), pp. 2271-2274, 2014.
[39] R. Sun, Z. Wang, K. E. Martens and S. Lewis, “Convolutional 3D Attention Network for Video Based Freezing of Gait Recognition,” Proceedings of the Digital Image Computing: Techniques and Applications (DICTA), pp. 1-7, 2018.
[40] A. Krajushkina, S. Nomm, A. Toomela, K. Medijainen, E. Tamm, M. Vaske, D. Uvarov, H. Kahar, M. Nugis and P. Taba, “Gait Analysis Based Approach for Parkinson's Disease Modeling with Decision Tree Classifiers,” Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp. 3720-3725, 2018.
[41] F. Aydin and Z. Aslan, “Classification of Neurodegenerative Diseases using Machine Learning Methods,” International Journal of Intelligent Systems and Applications in Engineering, Vol. 5, No. 1, pp. 1-9, 2017.
[42] T. Chen and C. Guestrin, “XGBoost: A Scalable Tree Boosting System,” Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 785-794, 2016.
[43] I. B. Mustapha and F. Saeed, “Bioactive Molecule Prediction Using Extreme Gradient Boosting,” Multidisciplinary Digital Publishing Institute, Vol. 21, No. 8, pp. 1-11, 2016.
[44] G. Paragliola and A. Coronato, “sEMG-Based Tremor Severity Evaluation for Parkinson's Disease Using a Light-Weight CNN,” IEEE Access, Vol. 26, No. 4, pp. 637-641, 2018.
[45] Z. Zheng, Y. Cai and Y. Li “Oversampling Method for Imbalanced Classification,” Journal of the Computing and informatics, Vol. 34, No. 5, pp. 1017-1037, 2016.
[46] G. Yogev, N. Giladi, C. Peretz, S. Springer, E. Simon and J. Hausdorff, “Dual tasking, gait rhythmicity, and Parkinson's disease: Which aspects of gait are attention demanding?,” European Journal of Neuroscience, Vol. 22, No. 5, pp. 1248-1256, 2005.
[47] Y. Mittra and V. Rustagi, “Classification of Subjects with Parkinson's Disease Using Gait Data Analysis,” Proceedings of the International Conference on Automation and Computational Engineering (ICACE), pp. 84-89, 2018.