Layer recurrent neural network-based diagnosis of Parkinson’s disease using voice features

Zehra Karapinar Senturk

doi:10.1515/bmt-2022-0022

Article

Layer recurrent neural network-based diagnosis of Parkinson’s disease using voice features

Zehra Karapinar Senturk

Published/Copyright: June 3, 2022

Published by

Become an author with De Gruyter Brill

Submit Manuscript Author Information

From the journal Biomedical Engineering / Biomedizinische Technik Volume 67 Issue 4

Abstract

Parkinson’s disease (PD), a slow-progressing neurological disease, affects a large percentage of the world’s elderly population, and this population is expected to grow over the next decade. As a result, early detection is crucial for community health and the future of the globe in order to take proper safeguards and have a less arduous treatment procedure. Recent research has begun to focus on the motor system deficits caused by PD. Because practically most of the PD patients suffer from voice abnormalities, researchers working on automated diagnostic systems investigate vocal impairments. In this paper, we undertake extensive experiments with features extracted from voice signals. We propose a layer Recurrent Neural Network (RNN) based diagnosis for PD. To prove the efficiency of the model, different network models are compared. To the best of our knowledge, several neural network topologies, namely RNN, Cascade Forward Neural Networks (CFNN), and Feed Forward Neural Networks (FFNN), are used and compared for voice-based PD detection for the first time. In addition, the impacts of data normalization and feature selection (FS) are thoroughly examined. The findings reveal that normalization increases classifier performance and Laplacian-based FS outperforms. The proposed RNN model with 300 voice features achieves 99.74% accuracy.

Keywords: early diagnosis; machine learning; neural networks; Parkinson’s disease; RNN

Corresponding author: Zehra Karapinar Senturk, Computer Engineering Department, Faculty of Engineering, Duzce University, 81620, Duzce, Turkey, E-mail: zehrakarapinar@duzce.edu.tr, zehrakarapinar@gmail.com

Research funding: Funding not received for the study.
Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
Competing interests: Authors state no conflict of interest.
Informed consent: A public dataset was used in this study and therefore, informed consent is not applicable for this study.
Ethical approval: The data used in this research has previously been approved by Clinical Research Ethics Committee of Bahcesehir University, Turkey. The data was dominated for public research by Sakar et al. [33] in 2018.

Appendix: Experimental results for different network topologies and network types

Table A1:

Experimental results of RNN.

Experiment No		Hidden neurons	Normalization method	Accuracy	Conf. Mat.
1	RNN	5	–	95.3632
2	RNN	10	–
3	RNN	[5 5]	–	94.8404	167	14
3	RNN	[5 5]	–	94.8404	25	550
4	RNN	[10 5]	–	96.8228	175	7
4	RNN	[10 5]	–	96.8228	17	557
5	RNN	5	Range [−1, 1]	96.0193	171	9
5	RNN	5	Range [−1, 1]	96.0193	21	555
6	RNN	[5 5]	Range [−1, 1]	96.9544	178	9
6	RNN	[5 5]	Range [−1, 1]	96.9544	14	555
7	RNN	10	Range [−1, 1]	95.7579	176	16
7	RNN	10	Range [−1, 1]	95.7579	16	548
8	RNN	12	Range [−1, 1]	95.1018	162	7
8	RNN	12	Range [−1, 1]	95.1018	30	557
9	RNN	[10 5]	Range [−1, 1]	97.6105	183	9
9	RNN	[10 5]	Range [−1, 1]	97.6105	9	555
10	RNN	[10 10]	Range [−1, 1]	96.5526	176	10
10	RNN	[10 10]	Range [−1, 1]	96.5526	16	554
11	RNN	[5 10]	Range [−1, 1]	95.6228	168	9
11	RNN	[5 10]	Range [−1, 1]	95.6228	24	555
12	RNN	[5 5 5]	Range [−1, 1]	94.8246	165	12
12	RNN	[5 5 5]	Range [−1, 1]	94.8246	27	552
13	RNN	5	Z score	96.0193	171	9
13	RNN	5	Z score	96.0193	21	555
14	RNN	[5 5]	Z score	96.9544	178	9
14	RNN	[5 5]	Z score	96.9544	14	555
15	RNN	10	Z score	95.7579	176	16
15	RNN	10	Z score	95.7579	16	548
16	RNN	12	Z score	95.1018	162	7
16	RNN	12	Z score	95.1018	30	557
17	RNN	[10 5]	Z score	97.6105	183	9
17	RNN	[10 5]	Z score	97.6105	9	555
18	RNN	[10 10]	Z score	96.2895	168	4
18	RNN	[10 10]	Z score	96.2895	24	560
19	RNN	[5 10]	Z score	95.6228	168	9
19	RNN	[5 10]	Z score	95.6228	24	555
20	RNN	[5 5 5]	Z score	94.8246	165	12
20	RNN	[5 5 5]	Z score	94.8246	27	552
21	RNN	5	2-norm	96.0193	171	9
21	RNN	5	2-norm	96.0193	21	555
22	RNN	[5 5]	2-norm	96.9544	178	9
22	RNN	[5 5]	2-norm	96.9544	14	555
23	RNN	10	2-norm	95.7579	176	16
23	RNN	10	2-norm	95.7579	16	548
24	RNN	12	2-norm	95.1018	162	7
24	RNN	12	2-norm	95.1018	30	557
25	RNN	[10 5]	2-norm	97.6105	183	9
25	RNN	[10 5]	2-norm	97.6105	9	555
26	RNN	[10 10]	2-norm	96.2895	168	4
26	RNN	[10 10]	2-norm	96.2895	24	560
27	RNN	[5 10]	2-norm	95.6228	168	9
27	RNN	[5 10]	2-norm	95.6228	24	555
28	RNN	[5 5 5]	2-norm	94.8246	165	12
28	RNN	[5 5 5]	2-norm	94.8246	27	552
29	RNN	5	Center	95.3632	169	12
29	RNN	5	Center	95.3632	23	552
30	RNN	[5 5]	Center	94.8404	167	14
30	RNN	[5 5]	Center	94.8404	25	550
31	RNN	10	Center	94.5684	171	20
31	RNN	10	Center	94.5684	21	544
32	RNN	12	Center	94.3018	168	19
32	RNN	12	Center	94.3018	24	545
33	RNN	[10 5]	Center	96.8228	175	7
33	RNN	[10 5]	Center	96.8228	17	557
34	RNN	[10 10]	Center	95.7579	171	11
34	RNN	[10 10]	Center	95.7579	21	553
35	RNN	[5 10]	Center	93.2561	163	22
35	RNN	[5 10]	Center	93.2561	29	542
36	RNN	[5 5 5]	Center	95.7632	170	10
36	RNN	[5 5 5]	Center	95.7632	22	554
37	RNN	5	Scale	96.0193	171	9
37	RNN	5	Scale	96.0193	21	555
38	RNN	[5 5]	Scale	96.9544	178	9
38	RNN	[5 5]	Scale	96.9544	14	555
39	RNN	10	Scale	95.7579	176	16
39	RNN	10	Scale	95.7579	16	548
40	RNN	12	Scale	95.1018	162	7
40	RNN	12	Scale	95.1018	30	557
41	RNN	[10 5]	Scale	97.6105	183	9
41	RNN	[10 5]	Scale	97.6105	9	555
42	RNN	[10 10]	Scale	96.2895	168	4
42	RNN	[10 10]	Scale	96.2895	24	560
43	RNN	[5 10]	Scale	95.6228	168	9
43	RNN	[5 10]	Scale	95.6228	24	555
44	RNN	[5 5 5]	Scale	94.8246	165	12
44	RNN	[5 5 5]	Scale	94.8246	27	552

Table A2:

Experimental results of CFNN.

Experiment No		Hidden neurons	Normalization method	Accuracy	Conf. Mat.
1	CFNN	5	–	93.9088	172	26
1	CFNN	5	–	93.9088	20	538
2	CFNN	10	–	92.8474	171	33
2	CFNN	10	–	92.8474	21	531
3	CFNN	[5 5]	–	93.6263	174	30
3	CFNN	[5 5]	–	93.6263	18	534
4	CFNN	[10 10]	–	96.6842	181	14
4	CFNN	[10 10]	–	96.6842	11	550
5	CFNN	5	Range [−1, 1]	95.093	175	20
5	CFNN	5	Range [−1, 1]	95.093	17	544
6	CFNN	[5 5]	Range [−1, 1]	95.2263	175	19
6	CFNN	[5 5]	Range [−1, 1]	95.2263	17	545
7	CFNN	10	Range [−1, 1]	94.5649	173	22
7	CFNN	10	Range [−1, 1]	94.5649	19	542
8	CFNN	12	Range [−1, 1]	93.6333	176	32
8	CFNN	12	Range [−1, 1]	93.6333	16	532
9	CFNN	[10 5]	Range [−1, 1]	95.4912	171	13
9	CFNN	[10 5]	Range [−1, 1]	95.4912	21	551
10	CFNN	[10 10]	Range [−1, 1]	95.8825	176	15
10	CFNN	[10 10]	Range [−1, 1]	95.8825	16	549
11	CFNN	[5 10]	Range [−1, 1]	94.0351	170	23
11	CFNN	[5 10]	Range [−1, 1]	94.0351	22	541
12	CFNN	[5 5 5]	Range [−1, 1]	95.6246	175	16
12	CFNN	[5 5 5]	Range [−1, 1]	95.6246	17	548
13	CFNN	5	Z score	95.093	175	20
13	CFNN	5	Z score	95.093	17	544
14	CFNN	[5 5]	Z score	95.2263	175	19
14	CFNN	[5 5]	Z score	95.2263	17	545
15	CFNN	10	Z score	94.5649	173	22
15	CFNN	10	Z score	94.5649	19	542
16	CFNN	12	Z score	93.6333	176	32
16	CFNN	12	Z score	93.6333	16	532
17	CFNN	[10 5]	Z score	95.4912	171	13
17	CFNN	[10 5]	Z score	95.4912	21	551
18	CFNN	[10 10]	Z score	95.8825	176	15
18	CFNN	[10 10]	Z score	95.8825	16	549
19	CFNN	[5 10]	Z score	94.0351	170	23
19	CFNN	[5 10]	Z score	94.0351	22	541
20	CFNN	[5 5 5]	Z score	95.6246	175	16
20	CFNN	[5 5 5]	Z score	95.6246	17	548
21	CFNN	5	2-norm	95.093	175	20
21	CFNN	5	2-norm	95.093	17	544
22	CFNN	[5 5]	2-norm	95.2263	175	19
22	CFNN	[5 5]	2-norm	95.2263	17	545
23	CFNN	10	2-norm	94.5649	173	22
23	CFNN	10	2-norm	94.5649	19	542
24	CFNN	12	2-norm	93.6333	176	32
24	CFNN	12	2-norm	93.6333	16	532
25	CFNN	[10 5]	2-norm	95.4912	171	13
25	CFNN	[10 5]	2-norm	95.4912	21	551
26	CFNN	[10 10]	2-norm	95.8825	176	15
26	CFNN	[10 10]	2-norm	95.8825	16	549
27	CFNN	[5 10]	2-norm	94.0351	170	23
27	CFNN	[5 10]	2-norm	94.0351	22	541
28	CFNN	[5 5 5]	2-norm	95.6246	175	16
28	CFNN	[5 5 5]	2-norm	95.6246	17	548
29	CFNN	5	Center	93.9088	172	26
29	CFNN	5	Center	93.9088	20	538
30	CFNN	[5 5]	Center	93.6263	174	30
30	CFNN	[5 5]	Center	93.6263	18	534
31	CFNN	10	Center	92.8474	171	33
31	CFNN	10	Center	92.8474	21	531
32	CFNN	12	Center	94.7	175	23
32	CFNN	12	Center	94.7	17	541
33	CFNN	[10 5]	Center	91.2561	163	37
33	CFNN	[10 5]	Center	91.2561	29	527
34	CFNN	[10 10]	Center	96.6842	181	14
34	CFNN	[10 10]	Center	96.6842	11	550
35	CFNN	[5 10]	Center	95.6175	173	14
35	CFNN	[5 10]	Center	95.6175	19	550
36	CFNN	[5 5 5]	Center	93.9035	177	31
36	CFNN	[5 5 5]	Center	93.9035	15	533
37	CFNN	5	Scale	95.093	175	20
37	CFNN	5	Scale	95.093	17	544
38	CFNN	[5 5]	Scale	95.2263	175	19
38	CFNN	[5 5]	Scale	95.2263	17	545
39	CFNN	10	Scale	94.5649	173	22
39	CFNN	10	Scale	94.5649	19	542
40	CFNN	12	Scale	93.633	176	32
40	CFNN	12	Scale	93.633	16	532
41	CFNN	[10 5]	Scale	95.4912	171	13
41	CFNN	[10 5]	Scale	95.4912	21	551
42	CFNN	[10 10]	Scale	95.8825	176	15
42	CFNN	[10 10]	Scale	95.8825	16	549
43	CFNN	[5 10]	Scale	94.0351	170	23
43	CFNN	[5 10]	Scale	94.0351	22	541
44	CFNN	[5 5 5]	Scale	95.6246	175	16
44	CFNN	[5 5 5]	Scale	95.6246	17	548

Table A3:

Experimental results of FFNN.

Experiment No		Hidden neurons	Normalization method	Accuracy	Conf. Mat.
1	FFNN	5	–	92.7158	142	5
1	FFNN	5	–	92.7158	50	559
2	FFNN	10	–	94.707	162	10
2	FFNN	10	–	94.707	30	554
3	FFNN	[5 5]	–	95.0965	171	16
3	FFNN	[5 5]	–	95.0965	21	548
4	FFNN	5	Range [−1, 1]	93.8982	175	29
4	FFNN	5	Range [−1, 1]	93.8982	17	535
5	FFNN	[5 5]	Range [−1, 1]	96.8228	179	11
5	FFNN	[5 5]	Range [−1, 1]	96.8228	13	553
6	FFNN	10	Range [−1, 1]	95.8912	178	17
6	FFNN	10	Range [−1, 1]	95.8912	14	547
7	FFNN	12	Range [−1, 1]	96.0211	176	14
7	FFNN	12	Range [−1, 1]	96.0211	16	550
8	FFNN	[10 5]	Range [−1, 1]	95.7667	167	7
8	FFNN	[10 5]	Range [−1, 1]	95.7667	25	557
9	FFNN	[10 10]	Range [−1, 1]	94.7018	164	12
9	FFNN	[10 10]	Range [−1, 1]	94.7018	28	552
10	FFNN	[5 10]	Range [−1, 1]	94.0386	167	20
10	FFNN	[5 10]	Range [−1, 1]	94.0386	25	544
11	FFNN	[5 5 5]	Range [−1, 1]	94.9526	174	20
11	FFNN	[5 5 5]	Range [−1, 1]	94.9526	18	544
12	FFNN	5	Z score	93.8982	175	29
12	FFNN	5	Z score	93.8982	17	535
13	FFNN	[5 5]	Z score	96.8228	179	11
13	FFNN	[5 5]	Z score	96.8228	13	553
14	FFNN	10	Z score	95.8912	178	17
14	FFNN	10	Z score	95.8912	14	547
15	FFNN	12	Z score	96.0211	176	14
15	FFNN	12	Z score	96.0211	16	550
16	FFNN	[10 5]	Z score	95.7667	167	7
16	FFNN	[10 5]	Z score	95.7667	25	557
17	FFNN	[10 10]	Z score	94.7018	164	12
17	FFNN	[10 10]	Z score	94.7018	28	552
18	FFNN	[5 10]	Z score	94.0386	167	20
18	FFNN	[5 10]	Z score	94.0386	25	544
19	FFNN	[5 5 5]	Z score	94.9526	174	20
19	FFNN	[5 5 5]	Z score	94.9526	18	544
20	FFNN	5	2-norm	93.8982	175	29
20	FFNN	5	2-norm	93.8982	17	535
21	FFNN	[5 5]	2-norm	96.8228	179	11
21	FFNN	[5 5]	2-norm	96.8228	13	553
22	FFNN	10	2-norm	95.8912	178	17
22	FFNN	10	2-norm	95.8912	14	547
23	FFNN	12	2-norm	96.0211	176	14
23	FFNN	12	2-norm	96.0211	16	550
24	FFNN	[10 5]	2-norm	95.7667	167	7
24	FFNN	[10 5]	2-norm	95.7667	25	557
25	FFNN	[10 10]	2-norm	94.7018	164	12
25	FFNN	[10 10]	2-norm	94.7018	28	552
26	FFNN	[5 10]	2-norm	94.0386	167	20
26	FFNN	[5 10]	2-norm	94.0386	25	544
27	FFNN	[5 5 5]	2-norm	94.9526	174	20
27	FFNN	[5 5 5]	2-norm	94.9526	18	544
28	FFNN	5	Center	92.7158	142	5
28	FFNN	5	Center	92.7158	50	559
29	FFNN	[5 5]	Center	95.0965	171	16
29	FFNN	[5 5]	Center	95.0965	21	548
30	FFNN	10	Center	94.707	162	10
30	FFNN	10	Center	94.707	30	554
31	FFNN	12	Center	93.3684	150	8
31	FFNN	12	Center	93.3684	42	556
32	FFNN	[10 5]	Center	96.6789	176	9
32	FFNN	[10 5]	Center	96.6789	16	555
33	FFNN	[10 10]	Center	95.7614	169	9
33	FFNN	[10 10]	Center	95.7614	23	555
34	FFNN	[5 10]	Center	92.9825	159	20
34	FFNN	[5 10]	Center	92.9825	33	544
35	FFNN	[5 5 5]	Center	94.9526	174	20
35	FFNN	[5 5 5]	Center	94.9526	18	544
36	FFNN	5	Scale	93.8982	175	29
36	FFNN	5	Scale	93.8982	17	535
37	FFNN	[5 5]	Scale	96.8228	179	11
37	FFNN	[5 5]	Scale	96.8228	13	553
38	FFNN	10	Scale	95.8912	178	17
38	FFNN	10	Scale	95.8912	14	547
39	FFNN	12	Scale	96.0211	176	14
39	FFNN	12	Scale	96.0211	16	550
40	FFNN	[10 5]	Scale	95.7667	167	7
40	FFNN	[10 5]	Scale	95.7667	25	557
41	FFNN	[10 10]	Scale	94.7018	164	12
41	FFNN	[10 10]	Scale	94.7018	28	552
42	FFNN	[5 10]	Scale	94.0386	167	20
42	FFNN	[5 10]	Scale	94.0386	25	544
43	FFNN	[5 5 5]	Scale	94.9526	174	20
43	FFNN	[5 5 5]	Scale	94.9526	18	544

References

1. Jankovic, J. Parkinson’s disease: clinical features and diagnosis. J Neurol Neurosurg Psychiatr 2008;79:368–76. https://doi.org/10.1136/jnnp.2007.131045.Search in Google Scholar PubMed

2. Von Campenhausen, S, Bornschein, B, Wick, R, Bötzel, K, Sampaio, C, Poewe, W, et al.. Prevalence and incidence of Parkinson’s disease in Europe. Eur Neuropsychopharmacol 2005;15:473–90. https://doi.org/10.1016/j.euroneuro.2005.04.007.Search in Google Scholar PubMed

3. Dorsey, ER, Sherer, T, Okun, MS, Bloem, BR. The emerging evidence of the Parkinson pandemic. J Parkinsons Dis 2018;8:S3–8. https://doi.org/10.3233/JPD-181474.Search in Google Scholar PubMed PubMed Central

4. Yang, W, Hamilton, JL, Kopil, C, Beck, JC, Tanner, CM, Albin, RL, et al.. Current and projected future economic burden of Parkinson’s disease in the U.S. NPJ Parkinsons Dis 2020;6:1–9. https://doi.org/10.1038/s41531-020-0117-1.Search in Google Scholar PubMed PubMed Central

5. Solana-Lavalle, G, Rosas-Romero, R. Classification of PPMI MRI scans with voxel-based morphometry and machine learning to assist in the diagnosis of Parkinson’s disease. Comput Methods Progr Biomed 2021;198:105793. https://doi.org/10.1016/j.cmpb.2020.105793.Search in Google Scholar PubMed

6. Singh, G, Vadera, M, Samavedham, L, Lim, ECH. Machine learning-based framework for multi-class diagnosis of neurodegenerative diseases: a study on Parkinson’s disease. IFAC-PapersOnLine 2016;49:990–5. https://doi.org/10.1016/j.ifacol.2016.07.331.Search in Google Scholar

7. Salvatore, C, Cerasa, A, Castiglioni, I, Gallivanone, F, Augimeri, A, Lopez, M, et al.. Machine learning on brain MRI data for differential diagnosis of Parkinson’s disease and Progressive Supranuclear Palsy. J Neurosci Methods 2014;222:230–7. https://doi.org/10.1016/j.jneumeth.2013.11.016.Search in Google Scholar PubMed

8. Abdulhay, E, Arunkumar, N, Narasimhan, K, Vellaiappan, E, Venkatraman, V. Gait and tremor investigation using machine learning techniques for the diagnosis of Parkinson disease. Future Generat Comput Syst 2018;83:366–73. https://doi.org/10.1016/j.future.2018.02.009.Search in Google Scholar

9. Balaji, E, Brindha, D, Elumalai, VK, Umesh, K. Data-driven gait analysis for diagnosis and severity rating of Parkinson’s disease. Med Eng Phys 2021;91:54–64. https://doi.org/10.1016/j.medengphy.2021.03.005.Search in Google Scholar PubMed

10. Balaji, E, Brindha, D, Balakrishnan, R. Supervised machine learning based gait classification system for early detection and stage classification of Parkinson’s disease. Appl Soft Comput 2020;94:106494. https://doi.org/10.1016/j.asoc.2020.106494.Search in Google Scholar

11. Oh, SL, Hagiwara, Y, Raghavendra, U, Yuvaraj, R, Arunkumar, N, Murugappan, M, et al.. A deep learning approach for Parkinson’s disease diagnosis from EEG signals. Neural Comput Appl 2018;32:10927–33. https://doi.org/10.1007/s00521-018-3689-5.Search in Google Scholar

12. Lee, S, Hussein, R, Ward, R, Jane Wang, Z, McKeown, MJ. A convolutional-recurrent neural network approach to resting-state EEG classification in Parkinson’s disease. J Neurosci Methods 2021;361:109282. https://doi.org/10.1016/j.jneumeth.2021.109282.Search in Google Scholar PubMed PubMed Central

13. Maidan, I, Zifman, N, Hausdorff, JM, Giladi, N, Levy-Lamdan, O, Mirelman, A. A multimodal approach using TMS and EEG reveals neurophysiological changes in Parkinson’s disease. Park Relat Disord 2021;89:28–33. https://doi.org/10.1016/j.parkreldis.2021.06.018.Search in Google Scholar PubMed

14. Khare, SK, Bajaj, V, Acharya, UR. Detection of Parkinson’s disease using automated tunable Q wavelet transform technique with EEG signals. Biocybern Biomed Eng 2021;41:679–89. https://doi.org/10.1016/j.bbe.2021.04.008.Search in Google Scholar

15. Parisi, L, RaviChandran, N, Manaog, ML. Feature-driven machine learning to improve early diagnosis of Parkinson’s disease. Expert Syst Appl 2018;110:182–90. https://doi.org/10.1016/j.eswa.2018.06.003.Search in Google Scholar

16. Karapinar Senturk, Z. Early diagnosis of Parkinson’s disease using machine learning algorithms. Med Hypotheses 2020;138:109603. https://doi.org/10.1016/j.mehy.2020.109603.Search in Google Scholar PubMed

17. Sakar, BE, Serbes, G, Sakar, CO. Analyzing the effectiveness of vocal features in early telediagnosis of Parkinson’s disease. PLoS One 2017;12:e0182428. https://doi.org/10.1371/journal.pone.0182428.Search in Google Scholar PubMed PubMed Central

18. Sakar, BE, Isenkul, ME, Sakar, CO, Sertbas, A, Gurgen, F, Delil, S, et al.. Collection and analysis of a Parkinson speech dataset with multiple types of sound recordings. IEEE J Biomed Health Inform 2013;17:828–34. https://doi.org/10.1109/jbhi.2013.2245674.Search in Google Scholar

19. Tăuţan, AM, Ionescu, B, Santarnecchi, E. Artificial intelligence in neurodegenerative diseases: a review of available tools with a focus on machine learning techniques. Artif Intell Med 2021;117:102081. https://doi.org/10.1016/j.artmed.2021.102081.Search in Google Scholar PubMed

20. Khan, T. Parkinson’s disease assessment using speech anomalies: a review. 2014. Available from: https://scholar.google.com/scholar?q=Parkinsons%20disease%20assessment%20using%20speech%20anomalies:%20a%20review.Search in Google Scholar

21. Rusz, J, Tykalova, T, Novotny, M, Zogala, D, Sonka, K, Ruzicka, E, et al.. Defining speech subtypes in de novo Parkinson disease: response to long-term levodopa therapy. Neurology 2021;97:E2124–35. https://doi.org/10.1212/WNL.0000000000012878.Search in Google Scholar PubMed

22. Dawson, VL, Dawson, TM. Promising disease-modifying therapies for Parkinson’s disease. Sci Transl Med 2019;11:eaba1659. https://doi.org/10.1126/scitranslmed.aba1659.Search in Google Scholar PubMed

23. Rusz, J, Hlavnička, J, Novotný, M, Tykalová, T, Pelletier, A, Montplaisir, J, et al.. Speech biomarkers in rapid eye movement sleep behavior disorder and Parkinson disease. Ann Neurol 2021;90:62–75. https://doi.org/10.1002/ana.26085.Search in Google Scholar PubMed PubMed Central

24. Kuresan, H, Samiappan, D, Maneesh Reddy, P, Sai Mahesh, R, Sriharsha, K. Parkinson’s disease analysis using speech signal processing – a research inclination for graduates. Mater Today Proc 2021.10.1016/j.matpr.2020.12.1203Search in Google Scholar

25. Kara, S, Güven, A, Öner, AÖ. Utilization of artificial neural networks in the diagnosis of optic nerve diseases. Comput Biol Med 2006;36:428–37. https://doi.org/10.1016/j.compbiomed.2005.01.003.Search in Google Scholar PubMed

26. Hosseini, HG, Luo, D, Reynolds, KJ. The comparison of different feed forward neural network architectures for ECG signal diagnosis. Med Eng Phys 2006;28:372–8. https://doi.org/10.1016/j.medengphy.2005.06.006.Search in Google Scholar PubMed

27. Omiotek, Z, Burda, A, Wójcik, W. The use of decision tree induction and artificial neural networks for automatic diagnosis of Hashimoto’s disease. Expert Syst Appl 2013;40:6684–9. https://doi.org/10.1016/j.eswa.2013.03.022.Search in Google Scholar

28. Senturk, ZK, Kara, R. Breast cancer diagnosis via data mining: performance analysis of seven different algorithms. Comput Sci Eng 2014;4:35–46. https://doi.org/10.5121/cseij.2014.4104.Search in Google Scholar

29. Karapınar Şentürk, Z. Artificial neural networks based decision support system for the detection of diabetic retinopathy artificial neural networks based decision support system for the detection of diabetic retinopathy. Sakarya Univ J Sci 2020;24:424–31.10.16984/saufenbilder.630482Search in Google Scholar

30. Er, MB, Isik, E, Isik, I. Parkinson’s detection based on combined CNN and LSTM using enhanced speech signals with Variational mode decomposition. Biomed Signal Process Control 2021;70:103006. https://doi.org/10.1016/j.bspc.2021.103006.Search in Google Scholar

31. Hosseini-Kivanani, N, Camilo Vásquez-Correa, J, Stede, M, Nöth, E. Automated cross-language intelligibility analysis of Parkinson’s disease patients using speech recognition technologies. In: 57th Annual meeting of the Association for Computational Linguistics: Student Research Workshop. Florence: Association for Computational Linguistics; 2019.10.18653/v1/P19-2010Search in Google Scholar

32. Gürüler, H. A novel diagnosis system for Parkinson’s disease using complex-valued artificial neural network with k-means clustering feature weighting method. Neural Comput Appl 2017;28:1657–66.10.1007/s00521-015-2142-2Search in Google Scholar

33. Sakar, CO, Serbes, G, Gunduz, A, Tunc, HC, Nizam, H, Sakar, BE, et al.. A comparative analysis of speech signal processing algorithms for Parkinson’s disease classification and the use of the tunable Q-factor wavelet transform. Appl Soft Comput 2019;74:255–63. https://doi.org/10.1016/j.asoc.2018.10.022.Search in Google Scholar

34. Gunduz, H. Deep learning-based Parkinson’s disease classification using vocal feature sets. IEEE Access 2019;7:115540–51. https://doi.org/10.1109/access.2019.2936564.Search in Google Scholar

35. Goyal, J, Khandnor, P, Aseri, TC. A hybrid approach for Parkinson’s disease diagnosis with resonance and time-frequency based features from speech signals. Expert Syst Appl 2021;182:115283. https://doi.org/10.1016/j.eswa.2021.115283.Search in Google Scholar

36. UCI Machine Learning Repository: Parkinsons Data Set. Available at: https://archive.ics.uci.edu/ml/datasets/parkinsons.Search in Google Scholar

37. Rusz, J, Tykalová, T, Novotný, M, Zogala, D, Růžička, E, Dušek, P. Automated speech analysis in early untreated Parkinson’s disease: relation to gender and dopaminergic transporter imaging. Eur J Neurol 2022;29:81–90. https://doi.org/10.1111/ene.15099.Search in Google Scholar PubMed

38. Rusz, J, Tykalová, T, Novotný, M, Růžička, E, Dušek, P. Distinct patterns of speech disorder in early-onset and late-onset de-novo Parkinson’s disease. NPJ Parkinsons Dis 2021;7:1–8. https://doi.org/10.1038/s41531-021-00243-1.Search in Google Scholar PubMed PubMed Central

39. Boersma, P, Weenink, D. Praat: doing phonetics by computer. 2011. Available at: https://www.fon.hum.uva.nl/praat/.Search in Google Scholar

40. MathWorks. Audio Toolbox – MATLAB & Simulink. Available at: https://www.mathworks.com/products/audio.html.Search in Google Scholar

41. Peng, H, Long, F, Ding, C. Feature selection based on mutual information: criteria of max-dependency, max-relevance, and min-redundancy. IEEE Trans Pattern Anal Mach Intell 2005;27:1226–38. https://doi.org/10.1109/tpami.2005.159.Search in Google Scholar PubMed

42. Sakar, CO, Kursun, O, Gurgen, F. A feature selection method based on kernel canonical correlation analysis and the minimum redundancy–maximum relevance filter method. Expert Syst Appl 2012;39:3432–7. https://doi.org/10.1016/j.eswa.2011.09.031.Search in Google Scholar

43. Kaya, D. The mRMR-CNN based influential support decision system approach to classify EEG signals. Measurement 2020;156:107602. https://doi.org/10.1016/j.measurement.2020.107602.Search in Google Scholar

44. Eroğlu, Y, Yildirim, M, Çinar, A. Convolutional Neural Networks based classification of breast ultrasonography images by hybrid method with respect to benign, malignant, and normal using mRMR. Comput Biol Med 2021;133:104407. https://doi.org/10.1016/j.compbiomed.2021.104407.Search in Google Scholar PubMed

45. Yuan, F, Lu, L, Zhang, YH, Wang, SP, Cai, YD. Data mining of the cancer-related lncRNAs GO terms and KEGG pathways by using mRMR method. Math Biosci 2018;304:1–8. https://doi.org/10.1016/j.mbs.2018.08.001.Search in Google Scholar PubMed

46. Dogan, A, Akay, M, Barua, PD, Baygin, M, Dogan, S, Tuncer, T, et al.. PrimePatNet87: prime pattern and tunable q-factor wavelet transform techniques for automated accurate EEG emotion recognition. Comput Biol Med 2021;138:104867. https://doi.org/10.1016/j.compbiomed.2021.104867.Search in Google Scholar PubMed

47. Pashaei, E, Pashaei, E. Gene selection using hybrid dragonfly black hole algorithm: a case study on RNA-seq COVID-19 data. Anal Biochem 2021;627:114242. https://doi.org/10.1016/j.ab.2021.114242.Search in Google Scholar PubMed

48. He, X, Cai, D, Niyogi, P. Laplacian score for feature selection. In: NIPS Proceedings 2005.Search in Google Scholar

49. Barile, C, Casavola, C, Pappalettera, G, Paramsamy Kannan, V. Laplacian score and K-means data clustering for damage characterization of adhesively bonded CFRP composites by means of acoustic emission technique. Appl Acoust 2022;185:108425. https://doi.org/10.1016/j.apacoust.2021.108425.Search in Google Scholar

50. Mohd Saufi, MSR, Hassan, KA. Remaining useful life prediction using an integrated Laplacian-LSTM network on machinery components. Appl Soft Comput 2021;112:107817. https://doi.org/10.1016/j.asoc.2021.107817.Search in Google Scholar

51. Yan, X, Liu, Y, Huang, D, Jia, M. A new approach to health condition identification of rolling bearing using hierarchical dispersion entropy and improved Laplacian score. Struct Health Monit 2020;20:1169–95. https://doi.org/10.1177/1475921720948620.Search in Google Scholar

52. Elman, JL. Finding structure in time. Cognit Sci 1990;14:179–211. https://doi.org/10.1207/s15516709cog1402_1.Search in Google Scholar

53. Ribeiro, LCF, Afonso, LCS, Papa, JP. Bag of samplings for computer-assisted Parkinson’s disease diagnosis based on recurrent neural networks. Comput Biol Med 2019;115:103477. https://doi.org/10.1016/j.compbiomed.2019.103477.Search in Google Scholar PubMed

54. Öztemel, E. Yapay Sinir Ağları, 4th ed. İstanbul: Papatya Bilim; 2016.Search in Google Scholar

55. Hogan, CA, Rajpurkar, P, Sowrirajan, H, Phillips, NA, Le, AT, Wu, M, et al.. Nasopharyngeal metabolomics and machine learning approach for the diagnosis of influenza. EBioMedicine 2021;71:103546. https://doi.org/10.1016/j.ebiom.2021.103546.Search in Google Scholar PubMed PubMed Central

56. Skodda, S, Visser, W, Schlegel, U. Gender-related patterns of dysprosody in Parkinson disease and correlation between speech variables and motor symptoms. J Voice 2011;25:76–82. https://doi.org/10.1016/j.jvoice.2009.07.005.Search in Google Scholar PubMed

57. Rusz, J, Novotny, M, Hlavnicka, J, Tykalova, T, Ruzicka, E. 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 Trans Neural Syst Rehabil Eng 2017;25:1319–21. https://doi.org/10.1109/tnsre.2016.2621885.Search in Google Scholar PubMed

58. Suphinnapong, P, Phokaewvarangkul, O, Thubthong, N, Teeramongkonrasmee, A, Mahattanasakul, P, Lorwattanapongsa, P, et al.. Objective vowel sound characteristics and their relationship with motor dysfunction in Asian Parkinson’s disease patients. J Neurol Sci 2021;426:117487. https://doi.org/10.1016/j.jns.2021.117487.Search in Google Scholar PubMed

59. Rusz, J, Tykalova, T, Ramig, LO, Tripoliti, E. Guidelines for speech recording and acoustic analyses in dysarthrias of movement disorders. Mov Disord 2021;36:803–14. https://doi.org/10.1002/mds.28465.Search in Google Scholar PubMed

60. Rusz, J, Švihlík, J, Krýže, P, Novotný, M, Tykalová, T. Reproducibility of voice analysis with machine learning. Mov Disord 2021;36:1282–3. https://doi.org/10.1002/mds.28604.Search in Google Scholar PubMed

Received: 2022-01-13

Accepted: 2022-05-18

Published Online: 2022-06-03

Published in Print: 2022-08-26

You are currently not able to access this content.

Articles in the same Issue

https://doi.org/10.1515/bmt-2022-0022

Keywords for this article

early diagnosis; machine learning; neural networks; Parkinson’s disease; RNN