du.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • chicago-author-date
  • chicago-note-bibliography
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Treatment response index from a multi-modal sensor fusion platform for assessment of motor states in Parkinson's disease
Dalarna University, School of Technology and Business Studies, Microdata Analysis.ORCID iD: 0000-0002-1548-5077
Show others and affiliations
2019 (English)Manuscript (preprint) (Other academic)
Abstract [en]

The aim of this paper is to develop and evaluate a multi-sensor data fusion platform for quantifying Parkinson’s disease (PD) motor states. More specifically, the aim is to evaluate the clinimetric properties (validity, reliability, and responsiveness to treatment) of the method, using data from motion sensors during lower- and upper-limb tests.

Methods: Nineteen PD patients and 22 healthy controls were recruited in a single center study. Subjects performed standardized motor tasks of Unified PD Rating Scale (UPDRS), including leg agility, hand rotation, and walking after wearing motion sensors on ankles and wrists. PD patients received a single levodopa dose before and at follow-up time points after the dose administration. Patients were video recorded and their motor symptoms were rated by three movement disorder experts. Experts rated each and every test occasions based on the six items of UPDRS-III (motor section), the treatment response scale (TRS) and the dyskinesia score. Spatiotemporal features were extracted from the sensor data. Features from lower limbs and upper limbs were fused. Feature selection methods of stepwise regression (SR), Lasso regression and principle component analysis (PCA) were used to select the most important features. Different machine learning methods of linear regression (LR), decision trees, and support vector machines were examined and their clinimetric properties were assessed.

Results: Treatment response index from multimodal motion sensors (TRIMMS) scores obtained from the most valid method of LR when using data from all tests. Features were selected by SR, and this method resulted in r=0.95 to TRS. The test-retest reliability of TRIMMS was good with intra-class correlation coefficient of 0.82. Responsiveness of the TRIMMS to levodopa treatment was similar to the responsiveness of TRS.

Conclusions: The results from this study indicate that fusing motion sensors data gathered during standardized motor tasks leads to valid, reliable and sensitive objective measurements of PD motor symptoms. These measurements could be further utilized in studies for individualized optimization of treatments in PD.

Place, publisher, year, edition, pages
2019.
National Category
Medical Engineering
Research subject
Complex Systems – Microdata Analysis
Identifiers
URN: urn:nbn:se:du-29546OAI: oai:DiVA.org:du-29546DiVA, id: diva2:1290688
Available from: 2019-02-21 Created: 2019-02-21 Last updated: 2019-02-27Bibliographically approved
In thesis
1. Multisensor data-driven methods for automated quantification of motor symptoms in Parkinson’s disease
Open this publication in new window or tab >>Multisensor data-driven methods for automated quantification of motor symptoms in Parkinson’s disease
2019 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

The overall aim of this thesis was to develop and evaluate new data-driven methods for supporting treatment and providing information for better management of Parkinson’s disease (PD).

This disease is complex and progressive. There is a large amount of inter- and intra-variability in motor symptoms of patients with PD (PwPD). Current evaluation of motor symptoms which is done at clinics by using clinical rating scales provides limited and only part of the health status of PwPD. PD requires an accurate assessment that is approved by clinics. Therefore frequent evaluation of symptoms at micro-level is required.

Sensor systems including smartphone and motion sensors were employed to collect data from PwPD and the recruited healthy controls. Repeated measures consisting of subjective assessment of symptoms and objective assessment of motor functions were collected.

First, the smartphone-based data-driven methods were developed to quantify the dexterity presented in fine motor tests of spiral drawing and alternate tapping. The upper extremities temporal irregularity measure presented in spiral drawing tests of PwPD was further analyzed by the approximate entropy (ApEn) method. Second, tri-axial motion sensor data were collected from various tests like leg agility, walking, and rapid alternating movements of hands of PwPD during a full cycled levodopa challenge. Data driven methods for quantification of leg agility tests and a combination of multiple motor tests were developed. The clinimetric properties of the methods such as reliability, validity, and responsiveness were evaluated. In addition, the feasibility of using smartphone inertial measurement unit (IMU) sensors in comparison to motion sensors for quantifying the motor states in PD during rapid alternating movements of hands tests was investigated.

Results of the developed methods for quantification of PD motor symptoms via dexterity tests in a smartphone can be used for measuring treatment related changes in PwPD. Investigation of the ApEn method showed good sensitivity and test-retest reliability indicating that this method is useful in measuring upper limb temporal irregularity at micro-level. High convergence validity resulted from using motion sensors during leg agility tests which led to valid and reliable objective measures of PD motor symptoms. The results of fusion of sensor data gathered during standardized motor tests were promising and led to highly valid, reliable and sensitive objective measures of PD motor symptoms. The results of the analyzing acceleration IMU data showed that smartphone IMU is capable of capturing symptom information from hand rotation tests. It can provide sufficient data for quantification of the motor states.

The findings from the data-driven methodology in this thesis can be used in development of systems for follow up of the effects of treatment and individualizing treatments in PD.

Place, publisher, year, edition, pages
Borlänge: Dalarna University, 2019
Series
Dalarna Doctoral Dissertations ; 10
Keywords
Parkinson’s disease, motor symptoms, motion sensors, smartphone, microdata, multivariate analysis, data-driven, support vector machine, stepwise regression, predictive models
National Category
Medical Laboratory and Measurements Technologies Biomedical Laboratory Science/Technology Computer Sciences
Research subject
Complex Systems – Microdata Analysis
Identifiers
urn:nbn:se:du-29583 (URN)978-91-88679-00-0 (ISBN)
Public defence
2019-04-26, Clas Ohlson, Borlänge, 13:00 (English)
Opponent
Supervisors
Available from: 2019-03-28 Created: 2019-02-27 Last updated: 2019-03-28Bibliographically approved

Open Access in DiVA

No full text in DiVA

Authority records BETA

Aghanavesi, Somayeh

Search in DiVA

By author/editor
Aghanavesi, Somayeh
By organisation
Microdata Analysis
Medical Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 215 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • chicago-author-date
  • chicago-note-bibliography
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf