Deep-deep neural network language models for predicting mild cognitive impairment

Sylvester Olubolu Orimaye; Jojo Sze-Meng Wong; Judyanne Sharmini Gilbert Fernandez

Deep-deep neural network language models for predicting mild cognitive impairment

Sylvester Olubolu Orimaye, Jojo Sze-Meng Wong, Judyanne Sharmini Gilbert Fernandez

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

Abstract

Early diagnosis of Mild Cognitive Impairment (MCI) is currently a challenge. Currently, MCI is diagnosed using specific clinical diagnostic criteria and neuropsychological examinations. As such we propose an automated diagnostic technique using a variant of deep neural networks language models (DNNLM) on the verbal utterances of MCI patients. Motivated by the success of DNNLM on natural language tasks, we propose a combination of deep neural network and deep language models (D2NNLM) to predict MCI. Results on the DementiaBank language transcript clinical dataset show that D2NNLM sufficiently learned several linguistic biomarkers in the form of higher order n-grams and skip-grams to distinguish the MCI group from the healthy group with reasonable accuracy, which could help clinical diagnosis even in the absence of sufficient training data.

Original language	English
Title of host publication	Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016)
Subtitle of host publication	co-located with 25th International Joint Conference on Artificial Intelligence (IJCAI 2016), New-York, USA, July 11, 2016 [proceedings]
Editors	Abdoulaye Baniré Diallo, Engelbert Mephu Nguifo, Mohammed Zaki
Publisher	Rheinisch-Westfaelische Technische Hochschule Aachen
Pages	14-20
Number of pages	7
Publication status	Published - 2016
Event	Advances in Bioinformatics and Artificial Intelligence 2016 - New York, United States of America Duration: 11 Jul 2016 → 11 Jul 2016 http://ceur-ws.org/Vol-1718/ (Proceedings)

Publication series

Name	CEUR Workshop Proceedings
Publisher	Rheinisch-Westfaelische Technische Hochschule Aachen Lehrstuhl Informatik V
Volume	1718
ISSN (Electronic)	1613-0073

Workshop

Workshop	Advances in Bioinformatics and Artificial Intelligence 2016
Abbreviated title	BAI 2016
Country	United States of America
City	New York
Period	11/07/16 → 11/07/16
Internet address	http://ceur-ws.org/Vol-1718/ (Proceedings)

Keywords

Deep neural networks
Mild cognitive impairment
Language models

Access to Document

33966488-oaFinal published version, 472 KBLicence: CC0

Cite this

Orimaye, S. O., Wong, J. S-M., & Gilbert Fernandez, J. S. (2016). Deep-deep neural network language models for predicting mild cognitive impairment. In A. Baniré Diallo, E. M. Nguifo, & M. Zaki (Eds.), Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016): co-located with 25th International Joint Conference on Artificial Intelligence (IJCAI 2016), New-York, USA, July 11, 2016 [proceedings] (pp. 14-20). (CEUR Workshop Proceedings; Vol. 1718). Rheinisch-Westfaelische Technische Hochschule Aachen. http://ceur-ws.org/Vol-1718/paper2.pdf

Orimaye, Sylvester Olubolu ; Wong, Jojo Sze-Meng ; Gilbert Fernandez, Judyanne Sharmini. / Deep-deep neural network language models for predicting mild cognitive impairment. Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016): co-located with 25th International Joint Conference on Artificial Intelligence (IJCAI 2016), New-York, USA, July 11, 2016 [proceedings]. editor / Abdoulaye Baniré Diallo ; Engelbert Mephu Nguifo ; Mohammed Zaki. Rheinisch-Westfaelische Technische Hochschule Aachen, 2016. pp. 14-20 (CEUR Workshop Proceedings).

@inproceedings{cd1b53f88e3347c1ae11296eea50ef22,

title = "Deep-deep neural network language models for predicting mild cognitive impairment",

abstract = "Early diagnosis of Mild Cognitive Impairment (MCI) is currently a challenge. Currently, MCI is diagnosed using specific clinical diagnostic criteria and neuropsychological examinations. As such we propose an automated diagnostic technique using a variant of deep neural networks language models (DNNLM) on the verbal utterances of MCI patients. Motivated by the success of DNNLM on natural language tasks, we propose a combination of deep neural network and deep language models (D2NNLM) to predict MCI. Results on the DementiaBank language transcript clinical dataset show that D2NNLM sufficiently learned several linguistic biomarkers in the form of higher order n-grams and skip-grams to distinguish the MCI group from the healthy group with reasonable accuracy, which could help clinical diagnosis even in the absence of sufficient training data.",

keywords = "Deep neural networks, Mild cognitive impairment, Language models",

author = "Orimaye, {Sylvester Olubolu} and Wong, {Jojo Sze-Meng} and {Gilbert Fernandez}, {Judyanne Sharmini}",

year = "2016",

language = "English",

series = "CEUR Workshop Proceedings",

publisher = "Rheinisch-Westfaelische Technische Hochschule Aachen",

pages = "14--20",

editor = "{Banir{\'e} Diallo}, {Abdoulaye } and Nguifo, {Engelbert Mephu } and Zaki, {Mohammed }",

booktitle = "Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016)",

note = "Advances in Bioinformatics and Artificial Intelligence 2016, BAI 2016 ; Conference date: 11-07-2016 Through 11-07-2016",

url = "http://ceur-ws.org/Vol-1718/",

}

Orimaye, SO, Wong, JS-M & Gilbert Fernandez, JS 2016, Deep-deep neural network language models for predicting mild cognitive impairment. in A Baniré Diallo, EM Nguifo & M Zaki (eds), Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016): co-located with 25th International Joint Conference on Artificial Intelligence (IJCAI 2016), New-York, USA, July 11, 2016 [proceedings]. CEUR Workshop Proceedings, vol. 1718, Rheinisch-Westfaelische Technische Hochschule Aachen, pp. 14-20, Advances in Bioinformatics and Artificial Intelligence 2016, New York, United States of America, 11/07/16. <http://ceur-ws.org/Vol-1718/paper2.pdf>

Deep-deep neural network language models for predicting mild cognitive impairment. / Orimaye, Sylvester Olubolu; Wong, Jojo Sze-Meng; Gilbert Fernandez, Judyanne Sharmini.

Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016): co-located with 25th International Joint Conference on Artificial Intelligence (IJCAI 2016), New-York, USA, July 11, 2016 [proceedings]. ed. / Abdoulaye Baniré Diallo; Engelbert Mephu Nguifo; Mohammed Zaki. Rheinisch-Westfaelische Technische Hochschule Aachen, 2016. p. 14-20 (CEUR Workshop Proceedings; Vol. 1718).

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - Deep-deep neural network language models for predicting mild cognitive impairment

AU - Orimaye, Sylvester Olubolu

AU - Wong, Jojo Sze-Meng

AU - Gilbert Fernandez, Judyanne Sharmini

PY - 2016

Y1 - 2016

N2 - Early diagnosis of Mild Cognitive Impairment (MCI) is currently a challenge. Currently, MCI is diagnosed using specific clinical diagnostic criteria and neuropsychological examinations. As such we propose an automated diagnostic technique using a variant of deep neural networks language models (DNNLM) on the verbal utterances of MCI patients. Motivated by the success of DNNLM on natural language tasks, we propose a combination of deep neural network and deep language models (D2NNLM) to predict MCI. Results on the DementiaBank language transcript clinical dataset show that D2NNLM sufficiently learned several linguistic biomarkers in the form of higher order n-grams and skip-grams to distinguish the MCI group from the healthy group with reasonable accuracy, which could help clinical diagnosis even in the absence of sufficient training data.

AB - Early diagnosis of Mild Cognitive Impairment (MCI) is currently a challenge. Currently, MCI is diagnosed using specific clinical diagnostic criteria and neuropsychological examinations. As such we propose an automated diagnostic technique using a variant of deep neural networks language models (DNNLM) on the verbal utterances of MCI patients. Motivated by the success of DNNLM on natural language tasks, we propose a combination of deep neural network and deep language models (D2NNLM) to predict MCI. Results on the DementiaBank language transcript clinical dataset show that D2NNLM sufficiently learned several linguistic biomarkers in the form of higher order n-grams and skip-grams to distinguish the MCI group from the healthy group with reasonable accuracy, which could help clinical diagnosis even in the absence of sufficient training data.

KW - Deep neural networks

KW - Mild cognitive impairment

KW - Language models

UR - http://ceur-ws.org/Vol-1718/

M3 - Conference Paper

T3 - CEUR Workshop Proceedings

SP - 14

EP - 20

BT - Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016)

A2 - Baniré Diallo, Abdoulaye

A2 - Nguifo, Engelbert Mephu

A2 - Zaki, Mohammed

PB - Rheinisch-Westfaelische Technische Hochschule Aachen

T2 - Advances in Bioinformatics and Artificial Intelligence 2016

Y2 - 11 July 2016 through 11 July 2016

ER -

Orimaye SO, Wong JS-M, Gilbert Fernandez JS. Deep-deep neural network language models for predicting mild cognitive impairment. In Baniré Diallo A, Nguifo EM, Zaki M, editors, Second international workshop on Advances in Bioinformatics and Artificial Intelligence: Bridging the Gap (BAI 2016): co-located with 25th International Joint Conference on Artificial Intelligence (IJCAI 2016), New-York, USA, July 11, 2016 [proceedings]. Rheinisch-Westfaelische Technische Hochschule Aachen. 2016. p. 14-20. (CEUR Workshop Proceedings).