基于语音特征的帕金森病与多系统萎缩帕金森型的早期鉴别诊断

doi:10.3969/j.issn.1006-9771.2025.10.014

《中国康复理论与实践》 ›› 2025, Vol. 31 ›› Issue (10): 1227-1233.doi: 10.3969/j.issn.1006-9771.2025.10.014

基于语音特征的帕金森病与多系统萎缩帕金森型的早期鉴别诊断

马凌燕¹, 曹杰², 陈仲略², 任康²(), 冯涛¹^,³()

1.首都医科大学附属北京天坛医院神经病学中心运动障碍性疾病科，北京市 100070
2.华中科技大学臻络科学神经系统疾病智能数字医疗技术中心，华中科技大学人工智能与自动化学院，湖北武汉市 430074
3.国家神经系统疾病临床医学研究中心，北京市 100070

收稿日期:2025-06-18 修回日期:2025-08-28 出版日期:2025-10-25 发布日期:2025-11-10
通讯作者: 冯涛，E-mail： happyft@sina.com；任康，E-mail： renkang@gyenno.com
作者简介:马凌燕(1985-)，女，汉族，山东青岛市人，博士，主任医师，主要研究方向：帕金森病及运动障碍病遗传、分型、脑网络等。
基金资助:
北京市自然科学基金项目(7232048)

Early differential diagnosis between Parkinson's disease and multiple system atrophy-Parkinsonism based on speech feature

MA Lingyan¹, CAO Jie², CHEN Zhonglüe², REN Kang²(), FENG Tao¹^,³()

1. Center for Movement Disorders, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, 100070, China
2. HUST-GYENNO CNS Intelligent Digital Medicine Technology Center, School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China
3. China National Clinical Research Center for Neurological Disease, Beijing 100070, China

Received:2025-06-18 Revised:2025-08-28 Published:2025-10-25 Online:2025-11-10
Contact: REN Kang, E-mail: happyft@sina.com; FENG Tao, E-mail: renkang@gyenno.com
Supported by:
Beijing Natural Science Foundation(7232048)

摘要/Abstract

摘要：

目的基于语音信号分析和人工智能相结合的无创方法，实现帕金森病与多系统萎缩帕金森型(MSA-P)早期自动化鉴别诊断。
方法 2023年7月至2025年2月，于首都医科大学附属北京天坛医院运动障碍性疾病科招募病程< 5年的MSA-P患者48例、帕金森病患者76例。设计11种语音任务范式，提取语音信号的声门、发声、构音、韵律、音系特征，以及基于表征学习提取的深度特征，通过数据驱动的方法筛选出最具鉴别力的特征，构建多种机器学习模型，实现对帕金森病与MSA-P患者的分类识别，选择鉴别效能最强的诊断模型。
结果逻辑回归模型表现最佳。对病程< 2年的早期患者，帕金森病与MSA-P间的分类准确率92.5%，精确率95.9%，召回率92.2%。在所有病程< 5年的患者中，逻辑回归模型准确率89.1%，精确率91.6%，召回率92.4%。即使使用单一语音范式提取的特征进行分析，诊断准确率也可达77.7%。
结论语音信号在帕金森病与MSA-P的早期鉴别诊断中具有重要的应用潜力。

关键词: 帕金森病, 多系统萎缩, 语音分析, 机器学习, 鉴别诊断, 早期诊断

Abstract:

Objective To develop an early automated differential diagnosis between Parkinson's disease (PD) and multiple system atrophy-Parkinsonism (MSA-P) using a non-invasive combination of voice signal analysis and artificial intelligence.
Methods From July, 2023 to February, 2025, a total of 48 MSA-P patients and 76 PD patients with a course of less than five years were recruited from Beijing Tiantan Hospital, Capital Medical University. Voice features, such as glottal, phonatory, articulatory, prosodic, phonological and representation learning-based features were extracted from eleven voice tasks. A data-driven approach was used to identify the most discriminative features, which were utilized to construct diagnostic models using a variety of machine learning models. The diagnostic model with the strongest discriminative efficiency was selected.
Results The logistic regression model showed the best performance. For early-stage patients with a course less than two years, the diagnostic accuracy, precision and recall rate between PD and MSA-P were 92.5%, 95.9% and 92.2%, respectively. For all the patients with a course less than five years, the logistic regression model achieved an accuracy of 89.1%, a precision of 91.6%, and a recall rate of 92.4%. Even when features extracted from a single speech paradigm were used for analysis, the diagnostic accuracy could still reach 77.7%.
Conclusion Voice signals analysis is potential in the early differential diagnosis of PD and MSA-P.

Key words: Parkinson's disease, multiple system atrophy, voice analysis, machine learning, differential diagnosis, early diagnosis

中图分类号:

R742.5

马凌燕, 曹杰, 陈仲略, 任康, 冯涛. 基于语音特征的帕金森病与多系统萎缩帕金森型的早期鉴别诊断[J]. 《中国康复理论与实践》, 2025, 31(10): 1227-1233.

MA Lingyan, CAO Jie, CHEN Zhonglüe, REN Kang, FENG Tao. Early differential diagnosis between Parkinson's disease and multiple system atrophy-Parkinsonism based on speech feature[J]. Chinese Journal of Rehabilitation Theory and Practice, 2025, 31(10): 1227-1233.

图/表 7

表1

表2

表3

表4

表5

表6

表7

参考文献 40

[1]	JELLINGER K A. Heterogeneity of multiple system atrophy: an update[J]. Biomedicines, 2022, 10(3): 599. doi: 10.3390/biomedicines10030599
[2]	SHIN H W, HONG S W, YOUN Y C. Clinical aspects of the differential diagnosis of Parkinson's disease and Parkinsonism[J]. J Clin Neurol (Seoul), 2022, 18(3): 259.
[3]	HSIAO J H T, TANGLAY O, LI A A, et al. Role of oligodendrocyte lineage cells in multiple system atrophy[J]. Cells, 2023, 12(5): 739. doi: 10.3390/cells12050739
[4]	WISEMAN J A, HALLIDAY G M, DIERIKS B V. Neuronal α-synuclein toxicity is the key driver of neurodegeneration in multiple system atrophy[J]. Brain, 2025, 148(7): 2306-2319. doi: 10.1093/brain/awaf030
[5]	OGAWA T, HATANO T, KAMAGATA K, et al. White matter and nigral alterations in multiple system atrophy-Parkinsonian type[J]. NPJ Parkinson Dis, 2021, 7(1): 96. doi: 10.1038/s41531-021-00236-0
[6]	HEIM B, KRISMER F, DE MARZI R, et al. Magnetic resonance imaging for the diagnosis of Parkinson's disease[J]. J Neural Transm, 2017, 124: 915-964. doi: 10.1007/s00702-017-1717-8 pmid: 28378231
[7]	STANKOVIC I, FANCIULLI A, SIDOROFF V, et al. A review on the clinical diagnosis of multiple system atrophy[J]. Cerebellum, 2023, 22(5): 825-839. doi: 10.1007/s12311-022-01453-w
[8]	SEKIYA H, TIPTON P W, KAWAZOE M, et al. Current landscape of clinical diagnosis in multiple system atrophy: a 15-year analysis from 2008 to 2022[J]. Neurology, 2024, 103(11): e210021. doi: 10.1212/WNL.0000000000210021
[9]	CHELBAN V, BOCCHETTA M, HASSANEIN S, et al. An update on advances in magnetic resonance imaging of multiple system atrophy[J]. J Neurol, 2019, 266: 1036-1045. doi: 10.1007/s00415-018-9121-3 pmid: 30460448
[10]	DEGUCHI K, IKEDA K, KUME K, et al. Significance of the hot-cross bun sign on T2*-weighted MRI for the diagnosis of multiple system atrophy[J]. J Neurol, 2015, 262: 1433-1439. doi: 10.1007/s00415-015-7728-1 pmid: 25845765
[11]	MARQUES T M, VAN RUMUND A, OECKL P, et al. Serum NFL discriminates Parkinson disease from atypical Parkinsonisms[J]. Neurology, 2019, 92(13): e1479-e1486.
[12]	MEI J, DESROSIERS C, FRASNELLI J. Machine learning for the diagnosis of Parkinson's disease: a review of literature[J]. Front Aging Neurosci, 2021, 13: 633752. doi: 10.3389/fnagi.2021.633752
[13]	ARYA A D, VERMA S S, CHAKARABARTI P, et al. A systematic review on machine learning and deep learning techniques in the effective diagnosis of Alzheimer's disease[J]. Brain Inform, 2023, 10(1): 17. doi: 10.1186/s40708-023-00195-7 pmid: 37450224
[14]	NOELLA R N, PRIYADARSHINI J. Machine learning algorithms for the diagnosis of Alzheimer and Parkinson disease[J]. J Med Eng Technol, 2023, 47(1): 35-43. doi: 10.1080/03091902.2022.2097326
[15]	TĂUŢAN A M, IONESCU B, SANTARNECCHI E. Artificial intelligence in neurodegenerative diseases: a review of available tools with a focus on machine learning techniques[J]. Artif Intell Med, 2021, 117: 102081. doi: 10.1016/j.artmed.2021.102081
[16]	KHALIQ F, OBERHAUSER J, WAKHLOO D, et al. Decoding degeneration: the implementation of machine learning for clinical detection of neurodegenerative disorders[J]. Neural Regen Res, 2023, 18(6): 1235-1242. doi: 10.4103/1673-5374.355982 pmid: 36453399
[17]	CHUDZIK A, ŚLEDZIANOWSKI A, PRZYBYSZEWSKI A W. Machine learning and digital biomarkers can detect early stages of neurodegenerative diseases[J]. Sensors, 2024, 24(5): 1572. doi: 10.3390/s24051572
[18]	AGGARWAL N, SAINI B, GUPTA S. Role of artificial intelligence techniques and neuroimaging modalities in detection of Parkinson's disease: a systematic review[J]. Cogn Comput, 2024, 16(4): 2078-2115. doi: 10.1007/s12559-023-10175-y
[19]	KAVUNGAL D, MAGALHÃES P, KUMAR S T, et al. Artificial intelligence-coupled plasmonic infrared sensor for detection of structural protein biomarkers in neurodegenerative diseases[J]. Sci Adv, 2023, 9(28): eadg9644.
[20]	GUPTA R, KUMARI S, SENAPATI A, et al. New era of artificial intelligence and machine learning-based detection, diagnosis, and therapeutics in Parkinson's disease[J]. Ageing Res Rev, 2023, 90: 102013. doi: 10.1016/j.arr.2023.102013
[21]	SRINIVASAN S, RAMADASS P, MATHIVANAN S K, et al. Detection of Parkinson disease using multiclass machine learning approach[J]. Sci Rep, 2024, 14(1): 13813. doi: 10.1038/s41598-024-64004-9 pmid: 38877028
[22]	STEFANO G B. Artificial intelligence as a tool for the diagnosis and treatment of neurodegenerative diseases[J]. Brain Sci, 2023, 13(6): 938. doi: 10.3390/brainsci13060938
[23]	PANG H, YU Z, YU H, et al. Multimodal striatal neuromarkers in distinguishing Parkinsonian variant of multiple system atrophy from idiopathic Parkinson's disease[J]. CNS Neurosci Ther, 2022, 28(12): 2172-2182. doi: 10.1111/cns.13959 pmid: 36047435
[24]	TOLOSA E, GARRIDO A, SCHOLZ S W, et al. Challenges in the diagnosis of Parkinson's disease[J]. Lancet Neurol, 2021, 20(5): 385-397. doi: 10.1016/S1474-4422(21)00030-2 pmid: 33894193
[25]	PRATIHAR R, SANKAR R. Advancements in Parkinson's disease diagnosis: a comprehensive survey on biomarker integration and machine learning[J]. Comput, 2024, 13(11): 293. doi: 10.3390/computers13110293
[26]	RAJPURKAR P, CHEN E, BANERJEE O, et al. AI in health and medicine[J]. Nat Med, 2022, 28(1): 31-38. doi: 10.1038/s41591-021-01614-0 pmid: 35058619
[27]	SUPPA A, ASCI F, SAGGIO G, et al. Voice analysis with machine learning: one step closer to an objective diagnosis of essential tremor[J]. Mov Disord, 2021, 36(6): 1401-1410. doi: 10.1002/mds.v36.6
[28]	DI CESARE M G, PERPETUINI D, CARDONE D, et al. Machine learning-assisted speech analysis for early detection of Parkinson's disease: a study on speaker diarization and classification techniques[J]. Sensors, 2024, 24(5): 1499. doi: 10.3390/s24051499
[29]	KHASKHOUSSY R, AYED Y B. Speech processing for early Parkinson's disease diagnosis: machine learning and deep learning-based approach[J]. Soc Netw Anal Min, 2022, 12(1): 73. doi: 10.1007/s13278-022-00905-9
[30]	SCHULTZ B G, TARIGOPPULA V S A, NOFFS G, et al. Automatic speech recognition in neurodegenerative disease[J]. Int J Speech Technol, 2021, 24(3): 771-779. doi: 10.1007/s10772-021-09836-w
[31]	RUSZ J, BONNET C, KLEMPIŘ J, et al. Speech disorders reflect differing pathophysiology in Parkinson's disease, progressive supranuclear palsy and multiple system atrophy[J]. J Neurol, 2015, 262: 992-1001. doi: 10.1007/s00415-015-7671-1 pmid: 25683763
[32]	DAOUDI K, DAS B, TYKALOVA T, et al. Speech acoustic indices for differential diagnosis between Parkinson's disease, multiple system atrophy and progressive supranuclear palsy[J]. NPJ Parkinson Dis, 2022, 8(1): 142. doi: 10.1038/s41531-022-00389-6
[33]	JERKIĆ L, PETROVIĆ-LAZIĆ M, VUKOVIĆ M. Speech disorders in Parkinson's disease-characteristics, assessment and treatment[J]. Med Pregl, 2021, 74(3-4): 106-111. doi: 10.2298/MPNS2104106J
[34]	POSTUMA R B, BERG D, STERN M, et al. MDS clinical diagnostic criteria for Parkinson's disease[J]. Mov Disord, 2015, 30(12): 1591-1601. doi: 10.1002/mds.v30.12
[35]	WENNING G K, STANKOVIC I, VIGNATELLI L, et al. The Movement Disorder Society criteria for the diagnosis of multiple system atrophy[J]. Mov Disord, 2022, 37(6): 1131-1148. doi: 10.1002/mds.v37.6
[36]	GOOLLA M, CHESHIRE W P, ROSS O A, et al. Diagnosing multiple system atrophy: current clinical guidance and emerging molecular biomarkers[J]. Front Neurol, 2023, 14: 1210220. doi: 10.3389/fneur.2023.1210220
[37]	RUSZ J, CMEJLA R, TYKALOVA T, et al. Imprecise vowel articulation as a potential early marker of Parkinson's disease: effect of speaking task[J]. J Acoust Soc Am, 2013, 134(3): 2171-2181. doi: 10.1121/1.4816541 pmid: 23967947
[38]	TYKALOVA T, RUSZ J, KLEMPIR J, et al. Distinct patterns of imprecise consonant articulation among Parkinson's disease, progressive supranuclear palsy and multiple system atrophy[J]. Brain Lang, 2017, 165: 1-9. doi: S0093-934X(16)30141-9 pmid: 27894006
[39]	HUH Y E, PARK J, SUH M K, et al. Differences in early speech patterns between Parkinson variant of multiple system atrophy and Parkinson's disease[J]. Brain Lang, 2015, 147: 14-20. doi: 10.1016/j.bandl.2015.04.007 pmid: 25997172
[40]	KAEGI G, BHATIA K P, TOLOSA E. The role of DAT-SPECT in movement disorders[J]. J Neurol Neurosurg Psychiatry, 2010, 81(1): 5-12. doi: 10.1136/jnnp.2008.157370 pmid: 20019219

组别	n	性别(男/女)/n	年龄/岁	病程/年
MSA-P	48	18/30	60.8 (55.9~66.4)	2.0 (1.9~3.0)
帕金森病	76	39/37	64.1 (55.5~69.2)	2.9 (1.8~3.9)
χ²/U值		1.739	1601.500	1594.500
P值		0.187	0.255	0.240

编号	范式
SP01	持续元音/a/
SP02	持续元音/o/
SP03	持续元音/i/
SP04	持续元音/u/
SP05	重复发音/papapa/
SP06	重复发音/tatata/
SP07	重复发音/kakaka/
SP08	交替发音/pataka/
SP09	短文朗读：《北风与太阳》
SP10	绕口令：四十四只石狮子
SP11	绕口令：一个大花碗扣着一只大花活蛤蟆

模型	准确率	精确率	召回率	加权F1	AUC
XGB	0.825	0.836	0.902	0.822	0.930
SVC	0.913	0.978	0.882	0.914	0.938
RF	0.838	0.828	0.941	0.832	0.956
LR	0.925	0.959	0.922	0.926	0.970

模型	准确率	精确率	召回率	加权F1	AUC
XGB	0.823	0.822	0.941	0.813	0.901
SVC	0.863	0.873	0.932	0.860	0.952
RF	0.840	0.836	0.949	0.832	0.912
LR	0.891	0.916	0.924	0.891	0.959

范式	总权重	贡献特征数
SP01	0.133	1
SP02	0.585	9
SP03	0.000	0
SP04	0.687	12
SP05	0.358	6
SP06	0.274	4
SP07	0.486	5
SP08	0.250	4
SP09	0.287	5
SP10	0.054	1
SP11	0.223	3

基于语音特征的帕金森病与多系统萎缩帕金森型的早期鉴别诊断

Early differential diagnosis between Parkinson's disease and multiple system atrophy-Parkinsonism based on speech feature

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 40

相关文章 15

Metrics

本文评价

推荐阅读 0

模型	准确率	精确率	召回率	加权F1	AUC
XGB	0.731	0.775	0.847	0.723	0.764
SVC	0.749	0.780	0.872	0.738	0.789
RF	0.720	0.759	0.856	0.707	0.765
LR	0.777	0.821	0.856	0.774	0.824

病程	帕金森病	MSA-P	U值	P值
< 1年	0.750	1.000	32.000	0.433
< 2年	0.667	1.393	374.500	< 0.001
< 3年	0.800	1.463	888.500	< 0.001
< 4年	0.845	1.458	1414.500	< 0.001
< 5年	0.856	1.396	2034.500	< 0.001

[1]	陆丹丹, 姚敬智, 李姿, 王柯文, 孙鑫廖, 陈建敏, 许建文. 2014年至2023年帕金森病物理治疗的文献计量学分析[J]. 《中国康复理论与实践》, 2025, 31(8): 906-913.
[2]	杜雯倩, 张旭, 陈继伟, 王兴, 朱昆. 运动干预对帕金森病冻结步态效果的Meta分析[J]. 《中国康复理论与实践》, 2025, 31(7): 781-789.
[3]	陈梦缘, 王秋琴, 徐语晨, 刘洁, 张馨悦, 陈菊萍, 徐桂华. 帕金森病疼痛相关研究的文献计量分析[J]. 《中国康复理论与实践》, 2024, 30(7): 797-803.
[4]	吕美玲, 王洁, 曾维斯, 温晓婷, 楚鑫. 虚拟现实技术对帕金森病患者认知功能和生活质量影响的Meta分析[J]. 《中国康复理论与实践》, 2024, 30(6): 648-656.
[5]	胥琪玲, 姜晓煜, 毕鸿雁. 近10年非侵入性脑刺激治疗帕金森病的文献计量分析[J]. 《中国康复理论与实践》, 2024, 30(6): 665-674.
[6]	葛颖, 赵沃娃, 张路, 舒璇, 李佳蔚, 刘颖. 帕金森病伴冻结步态患者的运动功能和生活质量[J]. 《中国康复理论与实践》, 2024, 30(3): 339-344.
[7]	王铭琛, 张文宇, 张贤祚, 臧万里. 经颅直流电刺激对帕金森病患者认知功能和生活质量影响的Meta分析[J]. 《中国康复理论与实践》, 2024, 30(2): 183-188.
[8]	王靖萱, 吕迪阳, 方伯言. 帕金森病步态异常非药物康复循证研究：基于ClinicalTrials.gov数据库分析[J]. 《中国康复理论与实践》, 2023, 29(7): 816-821.
[9]	吴佳玉, 孟德涛, 梁潇潇. 体力活动水平对帕金森病发病风险的影响：基于CHARLS数据的横断面分析[J]. 《中国康复理论与实践》, 2023, 29(10): 1135-1139.
[10]	朱旭,刘静,董泽萍,仇大伟. 基于表面肌电图手势动作意图识别的系统综述[J]. 《中国康复理论与实践》, 2022, 28(9): 1032-1038.
[11]	石曼欣妤,孟德涛,方伯言. 帕金森病康复研究进展的可视化分析[J]. 《中国康复理论与实践》, 2022, 28(9): 1060-1064.
[12]	王洁萍,彭娟,张斐雪,樊必双,胥方元. 肢体协调训练对早期原发性帕金森病患者肺功能的效果[J]. 《中国康复理论与实践》, 2022, 28(8): 966-971.
[13]	张博超,杨朝,郭立泉,陈静,熊大曦. 基于机器学习的慢性阻塞性肺疾病急性加重预测模型的研究[J]. 《中国康复理论与实践》, 2022, 28(6): 678-683.
[14]	谭茗丹,冯锭瑶,陈曦,刘汉军,李咏雪. 中文版构音障碍影响程度量表对帕金森病患者的信度和效度[J]. 《中国康复理论与实践》, 2022, 28(6): 696-703.
[15]	席晓明,毕鸿雁. 腹部扭转运动对帕金森病患者抑郁、便秘、运动症状和生活质量的效果[J]. 《中国康复理论与实践》, 2022, 28(2): 220-226.