Integrated and Enhanced Pipeline System to Support Spoken Language Analytics for Screening Neurocognitive Disorders

被引：1

作者：

Meng, Helen ^{[1
,2
,3
]}

Mak, Brian ^{[4
]}

Mak, Man-Wai ^{[5
]}

Fung, Helene ^{[6
]}

Gong, Xianmin ^{[2
,6
]}

Kwok, Timothy ^{[7
,8
]}

Liu, Xunying ^{[1
]}

Mok, Vincent ^{[10
,11
,12
,13
]}

Wong, Patrick ^{[14
,15
]}

Woo, Jean ^{[7
,9
]}

Wu, Xixin ^{[2
]}

Wong, Ka Ho ^{[1
]}

Xu, Sean Shensheng ^{[5
,16
]}

Zheng, Naijun ^{[1
]}

Huang, Ranzo ^{[4
]}

Kang, Jiawen ^{[1
]}

Ke, Xiaoquan ^{[5
]}

Li, Junan

Li, Jinchao ^{[1
,3
]}

Wang, Yi ^{[1
]}

机构：

[1] Chinese Univ Hong Kong, Dept Syst Engn & Engn Management, Hong Kong, Peoples R China

[2] Chinese Univ Hong Kong, Stanley Ho Big Data Decis Analyt Res Ctr, Hong Kong, Peoples R China

[3] Chinese Univ Hong Kong, Ctr Perceptual & Interact Intelligence, Hong Kong, Peoples R China

[4] Hong Kong Univ Sci & Technol, Dept Comp Sci & Engn, Hong Kong, Peoples R China

[5] Hong Kong Polytech Univ, Dept Elect & Informat Engn, Hong Kong, Peoples R China

[6] Chinese Univ Hong Kong, Dept Psychol, Hong Kong, Peoples R China

[7] Chinese Univ Hong Kong, Dept Med & Therapeut, Hong Kong, Peoples R China

[8] Chinese Univ Hong Kong, Jockey Club Ctr Osteoporosis Care & Control, Hong Kong, Peoples R China

[9] Chinese Univ Hong Kong, Jockey Club Inst Aging, Hong Kong, Peoples R China

[10] Chinese Univ Hong Kong, Div Neurol, Dept Med & Therapeut, Hong Kong, Peoples R China

[11] Chinese Univ Hong Kong, Margaret KL Cheung Res Ctr Management Parkinson, Hong Kong, Peoples R China

[12] Chinese Univ Hong Kong, Li Ka Shing Inst Hlth Sci, Hong Kong, Peoples R China

[13] Chinese Univ Hong Kong, Gerald Choa Neurosci Inst, Hong Kong, Peoples R China

[14] Chinese Univ Hong Kong, Dept Linguist & Modern Languages, Hong Kong, Peoples R China

[15] Chinese Univ Hong Kong, Brain & Mind Inst, Hong Kong, Peoples R China

[16] Shenzhen Univ, Sch Biomed Engn, Shenzhen, Peoples R China

来源：

INTERSPEECH 2023 | 2023年

关键词：

diarization; speech recognition; NCD detection; neurocognitive disorder; dementia; ALZHEIMERS-DISEASE; SPEECH; DIARIZATION; RECOGNITION;

D O I：

10.21437/Interspeech.2023-2249

中图分类号：

O42 [声学];

学科分类号：

070206 ; 082403 ;

摘要：

This paper presents an enhanced pipeline system for automated screening of neurocognitive disorders, e.g. Alzheimer's Disease (AD), using spoken language technologies. To ensure local relevance, the pipeline is applied to two-way interactions between clinical assessors and older adult participants in spoken Cantonese, the predominant language used in Hong Kong. The pipeline includes: (i) Speaker diarization using speaker-turn-aware scoring to capture the temporal structure of conversations. (ii) ASR using XLS-R wav2vec 2.0 models further pre-trained on Cantonese speech data and fine-tuned. (iii) Language modelling using RoBERTa with further fine-tuning. (iv) AD screening with neural network classification. A reference benchmark is obtained using the ADReSS corpus where no diarization is needed, and the partial pipeline attained a competitive detection accuracy of 87.5%.

引用

页码：1713 / 1717

页数：5

共 49 条

[1] [Anonymous], 2002, 7 INT C SPOK LANG PR
[2] A STUDY OF LANGUAGE FUNCTIONING IN ALZHEIMER PATIENTS
APPELL, J
KERTESZ, A
FISMAN, M
[J]. BRAIN AND LANGUAGE, 1982, 17 (01) : 73 - 91
[3] Babu A., 2022, P INTERSPEECH
[4] Baevski A., 2020, wav2vec 2.0: A framework for self-supervised learning of speech representations
[5] Balagopalan A., 2021, ARXIV
[6] Balagopalan A., 2020, P INTERSPEECH
[7] THE NATURAL-HISTORY OF ALZHEIMERS-DISEASE - DESCRIPTION OF STUDY COHORT AND ACCURACY OF DIAGNOSIS
BECKER, JT
BOLLER, F
LOPEZ, OL
SAXTON, J
MCGONIGLE, KL
MOOSSY, J
HANIN, I
WOLFSON, SK
DETRE, K
HOLLAND, A
GUR, D
LATCHAW, R
BRENNER, R
[J]. ARCHIVES OF NEUROLOGY, 1994, 51 (06) : 585 - 594
[8] Bredin H., 2020, P ICASSP
[9] Chan A.S., 2006, Hong Kong List Learning Test, V2nd
[10] Cheung R., 2004, J INT NEUROPSYCHOLOG

← 1 2 3 4 5 →