Motion sensor-based assessment on fall risk and Parkinson's disease severity: A deep multi-source multi-task learning (DMML) Approach

Shuo Yu, Hsinchun Chen, Randall Brown, Scott J Sherman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Senior citizens face many challenges to their independent living. Fall and Parkinson's disease (PD) are among the most life-threatening events or conditions. Precise and prompt motion sensor-based assessments for fall risk and PD severity can lead to timely interventions that can relieve or ultimately eliminate the threats. However, past studies focus on ad hoc feature engineering, which is subjective and less transferrable to related domains. In this work, we propose a Deep Multi-source Multi-task Learning (DMML) approach that provides an integrated framework for sensor-based condition risk and severity assessment. We develop Convolutional Neural Networks (CNN) to extract features from sensor signals. We collect timed up and go (TUG) test data at a neurology clinic to evaluate our model. Five sensors are attached to 22 PD patients to collect motion data throughout the test. Our model achieves F-measure of 0.940 for assessing fall risks, and RMSE of 0.060 for assessing PD severities, significantly outperforming the benchmark methods.

Original languageEnglish (US)
Title of host publicationProceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages174-179
Number of pages6
ISBN (Electronic)9781538653777
DOIs
StatePublished - Jul 24 2018
Event6th IEEE International Conference on Healthcare Informatics, ICHI 2018 - New York, United States
Duration: Jun 4 2018Jun 7 2018

Other

Other6th IEEE International Conference on Healthcare Informatics, ICHI 2018
CountryUnited States
CityNew York
Period6/4/186/7/18

Fingerprint

Parkinson Disease
Learning
Sensors
Independent Living
Benchmarking
Neurology
Neural networks

Keywords

  • Convolutional neural networks
  • Deep learning
  • Fall
  • Motion sensors
  • Multi-task learning
  • Parkinson's disease
  • Risk assessment

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Networks and Communications
  • Health Informatics

Cite this

Yu, S., Chen, H., Brown, R., & Sherman, S. J. (2018). Motion sensor-based assessment on fall risk and Parkinson's disease severity: A deep multi-source multi-task learning (DMML) Approach. In Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018 (pp. 174-179). [8419360] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICHI.2018.00027

Motion sensor-based assessment on fall risk and Parkinson's disease severity : A deep multi-source multi-task learning (DMML) Approach. / Yu, Shuo; Chen, Hsinchun; Brown, Randall; Sherman, Scott J.

Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 174-179 8419360.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Yu, S, Chen, H, Brown, R & Sherman, SJ 2018, Motion sensor-based assessment on fall risk and Parkinson's disease severity: A deep multi-source multi-task learning (DMML) Approach. in Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018., 8419360, Institute of Electrical and Electronics Engineers Inc., pp. 174-179, 6th IEEE International Conference on Healthcare Informatics, ICHI 2018, New York, United States, 6/4/18. https://doi.org/10.1109/ICHI.2018.00027
Yu S, Chen H, Brown R, Sherman SJ. Motion sensor-based assessment on fall risk and Parkinson's disease severity: A deep multi-source multi-task learning (DMML) Approach. In Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 174-179. 8419360 https://doi.org/10.1109/ICHI.2018.00027
Yu, Shuo ; Chen, Hsinchun ; Brown, Randall ; Sherman, Scott J. / Motion sensor-based assessment on fall risk and Parkinson's disease severity : A deep multi-source multi-task learning (DMML) Approach. Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 174-179
@inproceedings{aa529c062b014c50abaf4bd93d6db8fb,
title = "Motion sensor-based assessment on fall risk and Parkinson's disease severity: A deep multi-source multi-task learning (DMML) Approach",
abstract = "Senior citizens face many challenges to their independent living. Fall and Parkinson's disease (PD) are among the most life-threatening events or conditions. Precise and prompt motion sensor-based assessments for fall risk and PD severity can lead to timely interventions that can relieve or ultimately eliminate the threats. However, past studies focus on ad hoc feature engineering, which is subjective and less transferrable to related domains. In this work, we propose a Deep Multi-source Multi-task Learning (DMML) approach that provides an integrated framework for sensor-based condition risk and severity assessment. We develop Convolutional Neural Networks (CNN) to extract features from sensor signals. We collect timed up and go (TUG) test data at a neurology clinic to evaluate our model. Five sensors are attached to 22 PD patients to collect motion data throughout the test. Our model achieves F-measure of 0.940 for assessing fall risks, and RMSE of 0.060 for assessing PD severities, significantly outperforming the benchmark methods.",
keywords = "Convolutional neural networks, Deep learning, Fall, Motion sensors, Multi-task learning, Parkinson's disease, Risk assessment",
author = "Shuo Yu and Hsinchun Chen and Randall Brown and Sherman, {Scott J}",
year = "2018",
month = "7",
day = "24",
doi = "10.1109/ICHI.2018.00027",
language = "English (US)",
pages = "174--179",
booktitle = "Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Motion sensor-based assessment on fall risk and Parkinson's disease severity

T2 - A deep multi-source multi-task learning (DMML) Approach

AU - Yu, Shuo

AU - Chen, Hsinchun

AU - Brown, Randall

AU - Sherman, Scott J

PY - 2018/7/24

Y1 - 2018/7/24

N2 - Senior citizens face many challenges to their independent living. Fall and Parkinson's disease (PD) are among the most life-threatening events or conditions. Precise and prompt motion sensor-based assessments for fall risk and PD severity can lead to timely interventions that can relieve or ultimately eliminate the threats. However, past studies focus on ad hoc feature engineering, which is subjective and less transferrable to related domains. In this work, we propose a Deep Multi-source Multi-task Learning (DMML) approach that provides an integrated framework for sensor-based condition risk and severity assessment. We develop Convolutional Neural Networks (CNN) to extract features from sensor signals. We collect timed up and go (TUG) test data at a neurology clinic to evaluate our model. Five sensors are attached to 22 PD patients to collect motion data throughout the test. Our model achieves F-measure of 0.940 for assessing fall risks, and RMSE of 0.060 for assessing PD severities, significantly outperforming the benchmark methods.

AB - Senior citizens face many challenges to their independent living. Fall and Parkinson's disease (PD) are among the most life-threatening events or conditions. Precise and prompt motion sensor-based assessments for fall risk and PD severity can lead to timely interventions that can relieve or ultimately eliminate the threats. However, past studies focus on ad hoc feature engineering, which is subjective and less transferrable to related domains. In this work, we propose a Deep Multi-source Multi-task Learning (DMML) approach that provides an integrated framework for sensor-based condition risk and severity assessment. We develop Convolutional Neural Networks (CNN) to extract features from sensor signals. We collect timed up and go (TUG) test data at a neurology clinic to evaluate our model. Five sensors are attached to 22 PD patients to collect motion data throughout the test. Our model achieves F-measure of 0.940 for assessing fall risks, and RMSE of 0.060 for assessing PD severities, significantly outperforming the benchmark methods.

KW - Convolutional neural networks

KW - Deep learning

KW - Fall

KW - Motion sensors

KW - Multi-task learning

KW - Parkinson's disease

KW - Risk assessment

UR - http://www.scopus.com/inward/record.url?scp=85051110170&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85051110170&partnerID=8YFLogxK

U2 - 10.1109/ICHI.2018.00027

DO - 10.1109/ICHI.2018.00027

M3 - Conference contribution

AN - SCOPUS:85051110170

SP - 174

EP - 179

BT - Proceedings - 2018 IEEE International Conference on Healthcare Informatics, ICHI 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -