Education, Science, Technology, Innovation and Life
Open Access
Sign In
  1. Home
  2. Journals
  3. MEDS Clinical Medicine
  4. Vol 5, Issue 2, 2024

A study on Parkinson's disease diagnosis by random forest algorithm based on improved speech features

Download as PDF

DOI: 10.23977/medsc.2024.050224 | Downloads: 23 | Views: 182

Author(s)

Zhijun Li 1, Jingxuan He 1, Di Sun 2, Haixia Li 3

Affiliation(s)

1 North China University of Technology, Beijing, 100144, China
2 Graduate College, Beijing University of Traditional Chinese Medicine, Beijing, 100029, China
3 Guang'anmen Hospital, China Academy of Chinese Medical Sciences, Beijing, 100053, China

Corresponding Author

Haixia Li

ABSTRACT

Neurological disorders have a serious impact on human life worldwide. Parkinson's disease, also known as idiopathic or primary Parkinson's disease, is one of the most common neurological disorders. In recent years, research related to the link between Parkinson's disease and speech has received more and more attention, and many methods to process speech signals through algorithms and thus predict the prevalence of Parkinson's have been proposed, and most of the studies have diagnosed the prevalence of Parkinson's disease by employing the speech signals of the subjects, and the results have mostly been better, successfully responding to the link between speech and Parkinson's disease. In this article, the speech signals of the subjects (554 cases in total, of which 220 cases are healthy people and 314 cases are Parkinson's patients) are collected, processed, and 12 kinds of complex speech features are extracted from them.By comparing these 12 kinds of speech features of Parkinson's patients and healthy people, three main classes of features are selected from them, which are Fundamental Frequency Perturbation Jitter Class, Amplitude Perturbation Shimmer Class, and Harmonic Signal-to-Noise Ratio (HNR) class.The speech features of the subjects are trained and tested by neural network, and comparative experiments are made on XGBoost algorithm, svm algorithm, random forest algorithm, KNN algorithm in machine learning and DNN neural network and LSTM neural network in deep learning.It is found that Random Forest Algorithm and can effectively solve the neural network's over-fitting and the problem of low accuracy and recall, and very effectively distinguish between Parkinson's patients and healthy people, with an accuracy rate of 99.3% and a recall rate of 100%.

KEYWORDS

Parkinson's disease, machine learning, deep learning, neural networks, speech diagnosis, artificial intelligence

CITE THIS PAPER

Zhijun Li, Jingxuan He, Di Sun, Haixia Li, A study on Parkinson's disease diagnosis by random forest algorithm based on improved speech features. MEDS Clinical Medicine (2024) Vol. 5: 152-160. DOI: http://dx.doi.org/10.23977/medsc.2024.050224.

REFERENCES

[1] Connollybs, Langae. Pharmacological treatment of Parkinson disease: a review[J].Jama, 2014,311( 16): 1670-1683.
[2] Pereira Cr, Webersa, Hook C, et al.Deep learning-aided Parkinson’s disease diagnosis from handwritten dynamics [C]. Proceedings of the 29th SIBGRAPI Conference on Graphics, Sao Paulo: Patterns and Images (SIBGRAPI), IEEE, 2016: 340-346
[3] Berus L, Klancnik S,Brezocnik M.Classifying Parkinson's disease based on acoustic measures using artificial neural networks[J]. Sensors, 2019, 19(1): 1-15.
[4] Shi Hao-Bin.Research on the diagnosis of speech disorders in Parkinson's disease based on convolutional neural network [D]. Qinhuangdao: Yanshan University, 2017.
[5] Huang Fang-Liang, Xu Huan-Qing, Shen Tong-Ping, Jin Li, Yu Lei. A study on Parkinson's disease identification combining residual neural network and speech diagnosis [J]. Journal of Qilu University of Technology, 2022, 36(01): 36-43. DOI:10.16442/j.cnki.qlgydxxb.2022.01.006.
[6] Hagen Jaeger, Dhaval Trivedi and Michael Stadtschnitzer.(2019). King's College London (MDVR-KCL) Mobile device recordings in patients with early and late onset Parkinson's disease and healthy controls [Dataset]. Zenodo.https://doi.org/10.5281/zenodo.2867216
[7] Sakar, C.O., Serbes, G., Gunduz, A., Tunc, H.C., Nizam, H., Sakar, B.E., Tutuncu, M., Aydin, T., Isenkul, M.E. and Apaydin, H., 2018. Speech signal processing for Parkinson's disease classification A comparative analysis of algorithms and the use of adjustable Q-factor wavelet transform. Applied Soft Computing.
[8] Zhang Tao, Peipei Jiang, Zhang Yajuan, et al. Research on the analysis method of speech impairment in Parkinson's disease based on local statistics in time-frequency mixed domain [J]. Journal of Biomedical Engineering, 2021, 38(1): 21-29. 
[9] Xu Jing. Exploring the quantitative assessment method of hypokinetic dysarthria Parkinson's disease based on acoustic features [D]. Guangzhou: Jinan University, 2020. 
[10] Liu Feng. Diagnosis and prediction of Parkinson's disease based on hand mapping and speech [D]. Nanjing: Nanjing University of Posts and Telecommunications, 2019. 
[11] F. Huang, H. Xu, T. Shen and L. Jin, "Recognition of Parkinson's Disease Based on Residual Neural Network and Voice Diagnosis," 2021 IEEE 5th Information Technology ,Networking, Electronic and Automation Control Conference (ITNEC), Xi'an, China, 2021, pp. 381-386, doi: 10.1109/ITNEC52019.2021.9586915.

Subscription
E-Mail Alert
Most popular papers
Call for Special Issue
Publication Ethics & OA Statement
Plagiarism Policy
Contact Us
Downloads: 5016
Visits: 227746

Sponsors, Associates, and Links

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.