Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model

Amir Hossein Poorjam Alavijeh, Yordan P. Raykov, Reham Badawy, Jesper Rindom Jensen, Mads Græsbøll Christensen, Max A. Little

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

152 Downloads (Pure)

Abstract

The performance of voice-based systems for remote monitoring of Parkinson's disease is highly dependent on the degree of adherence of the recordings to the test protocols, which probe for specific symptoms. Identifying segments of the signal that adhere to the protocol assumptions is typically performed manually by experts. This process is costly, time consuming, and often infeasible for large-scale data sets. In this paper, we propose a method to automatically identify the segments of signals that violate the test protocol with a high accuracy. In our approach, the signal is first split into variable duration segments by fitting an infinite hidden Markov model (iHMM) to the frames of the signals in the mel-frequency cepstral domain. The complexity of the iHMM is capable of growing jointly with the data allowing us to infer a potentially large (asymptotically infinite) number of different phenomena segmented into different hidden states. Then, we identify the segments that adhere to the test protocol by applying a multinomial naive Bayes classifier to the state indicators of segments. The experimental results show that even by using a small amount of training data, we can achieve around 96% accuracy in identifying short-term protocol violations with a 0.2 s resolution.
Original languageEnglish
Title of host publicationIEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
PublisherIEEE
Publication statusAccepted/In press - 2019

Fingerprint

Hidden Markov models
Quality control
Monitoring
Classifiers

Keywords

  • Bayesian Nonparametric
  • Infinite HMM
  • Parkinson's Disease
  • Quality Control
  • Segmentation

Cite this

Alavijeh, A. H. P., Raykov, Y. P., Badawy, R., Jensen, J. R., Christensen, M. G., & Little, M. A. (Accepted/In press). Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) IEEE.
Alavijeh, Amir Hossein Poorjam ; Raykov, Yordan P. ; Badawy, Reham ; Jensen, Jesper Rindom ; Christensen, Mads Græsbøll ; Little, Max A. / Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019.
@inproceedings{2e93b4b28dd34c9396d23f3a8b667f55,
title = "Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model",
abstract = "The performance of voice-based systems for remote monitoring of Parkinson's disease is highly dependent on the degree of adherence of the recordings to the test protocols, which probe for specific symptoms. Identifying segments of the signal that adhere to the protocol assumptions is typically performed manually by experts. This process is costly, time consuming, and often infeasible for large-scale data sets. In this paper, we propose a method to automatically identify the segments of signals that violate the test protocol with a high accuracy. In our approach, the signal is first split into variable duration segments by fitting an infinite hidden Markov model (iHMM) to the frames of the signals in the mel-frequency cepstral domain. The complexity of the iHMM is capable of growing jointly with the data allowing us to infer a potentially large (asymptotically infinite) number of different phenomena segmented into different hidden states. Then, we identify the segments that adhere to the test protocol by applying a multinomial naive Bayes classifier to the state indicators of segments. The experimental results show that even by using a small amount of training data, we can achieve around 96{\%} accuracy in identifying short-term protocol violations with a 0.2 s resolution.",
keywords = "Bayesian Nonparametric, Infinite HMM, Parkinson's Disease, Quality Control, Segmentation",
author = "Alavijeh, {Amir Hossein Poorjam} and Raykov, {Yordan P.} and Reham Badawy and Jensen, {Jesper Rindom} and Christensen, {Mads Gr{\ae}sb{\o}ll} and Little, {Max A.}",
year = "2019",
language = "English",
booktitle = "IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)",
publisher = "IEEE",
address = "United States",

}

Alavijeh, AHP, Raykov, YP, Badawy, R, Jensen, JR, Christensen, MG & Little, MA 2019, Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model. in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE.

Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model. / Alavijeh, Amir Hossein Poorjam; Raykov, Yordan P.; Badawy, Reham; Jensen, Jesper Rindom; Christensen, Mads Græsbøll; Little, Max A.

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019.

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model

AU - Alavijeh, Amir Hossein Poorjam

AU - Raykov, Yordan P.

AU - Badawy, Reham

AU - Jensen, Jesper Rindom

AU - Christensen, Mads Græsbøll

AU - Little, Max A.

PY - 2019

Y1 - 2019

N2 - The performance of voice-based systems for remote monitoring of Parkinson's disease is highly dependent on the degree of adherence of the recordings to the test protocols, which probe for specific symptoms. Identifying segments of the signal that adhere to the protocol assumptions is typically performed manually by experts. This process is costly, time consuming, and often infeasible for large-scale data sets. In this paper, we propose a method to automatically identify the segments of signals that violate the test protocol with a high accuracy. In our approach, the signal is first split into variable duration segments by fitting an infinite hidden Markov model (iHMM) to the frames of the signals in the mel-frequency cepstral domain. The complexity of the iHMM is capable of growing jointly with the data allowing us to infer a potentially large (asymptotically infinite) number of different phenomena segmented into different hidden states. Then, we identify the segments that adhere to the test protocol by applying a multinomial naive Bayes classifier to the state indicators of segments. The experimental results show that even by using a small amount of training data, we can achieve around 96% accuracy in identifying short-term protocol violations with a 0.2 s resolution.

AB - The performance of voice-based systems for remote monitoring of Parkinson's disease is highly dependent on the degree of adherence of the recordings to the test protocols, which probe for specific symptoms. Identifying segments of the signal that adhere to the protocol assumptions is typically performed manually by experts. This process is costly, time consuming, and often infeasible for large-scale data sets. In this paper, we propose a method to automatically identify the segments of signals that violate the test protocol with a high accuracy. In our approach, the signal is first split into variable duration segments by fitting an infinite hidden Markov model (iHMM) to the frames of the signals in the mel-frequency cepstral domain. The complexity of the iHMM is capable of growing jointly with the data allowing us to infer a potentially large (asymptotically infinite) number of different phenomena segmented into different hidden states. Then, we identify the segments that adhere to the test protocol by applying a multinomial naive Bayes classifier to the state indicators of segments. The experimental results show that even by using a small amount of training data, we can achieve around 96% accuracy in identifying short-term protocol violations with a 0.2 s resolution.

KW - Bayesian Nonparametric

KW - Infinite HMM

KW - Parkinson's Disease

KW - Quality Control

KW - Segmentation

M3 - Article in proceeding

BT - IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

PB - IEEE

ER -

Alavijeh AHP, Raykov YP, Badawy R, Jensen JR, Christensen MG, Little MA. Quality Control of Voice Recordings in Remote Parkinson's Disease Monitoring using the Infinite Hidden Markov Model. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE. 2019