Abstract
An impaired cognitive ability is an important indicator of dementia-related disease. Accordingly, this paper utilizes 12 inertia-related features, 19 gait-related features, and 2 balance-related features to analyze the gait performance of subjects while performing single-task and dual-task walking tests and balance tests. The features most closely correlated with the cognitive ability of the subjects are extracted via a correlation analysis method and a sequential forward floating selection (SFFS) algorithm, respectively. The extracted features are then used to predict the cognitive assessment screening instrument (CASI) and mini mental state examination (MMSE) scores of the subjects using three different prediction models, namely a linear regression model, a nonlinear regression model, and a feedforward neural network (FNN) model. It is shown that the optimal prediction performance (i.e., prediction error = 7.89±5.86 for the CASI score and prediction error = 3.21±2.86 for the MMSE score) is obtained using the SFFS feature selection method and the FNN model. Overall, the results show that the feature selection and modeling methods proposed in this paper provide an accurate and objective means of evaluating the cognitive ability of individuals for dementia diagnosis purposes.
| Original language | English |
|---|---|
| Pages (from-to) | 3338-3350 |
| Number of pages | 13 |
| Journal | IEEE Sensors Journal |
| Volume | 18 |
| Issue number | 8 |
| DOIs | |
| Publication status | Published - 2018 Apr 15 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Instrumentation
- Electrical and Electronic Engineering
Cite this
}
Predictive Models for Evaluating Cognitive Ability in Dementia Diagnosis Applications Based on Inertia- and Gait-Related Parameters. / Wang, Wei Hsin; Hsu, Yu Liang; Chung, Pau-Choo; Pai, Ming-Chyi.
In: IEEE Sensors Journal, Vol. 18, No. 8, 15.04.2018, p. 3338-3350.Research output: Contribution to journal › Article
TY - JOUR
T1 - Predictive Models for Evaluating Cognitive Ability in Dementia Diagnosis Applications Based on Inertia- and Gait-Related Parameters
AU - Wang, Wei Hsin
AU - Hsu, Yu Liang
AU - Chung, Pau-Choo
AU - Pai, Ming-Chyi
PY - 2018/4/15
Y1 - 2018/4/15
N2 - An impaired cognitive ability is an important indicator of dementia-related disease. Accordingly, this paper utilizes 12 inertia-related features, 19 gait-related features, and 2 balance-related features to analyze the gait performance of subjects while performing single-task and dual-task walking tests and balance tests. The features most closely correlated with the cognitive ability of the subjects are extracted via a correlation analysis method and a sequential forward floating selection (SFFS) algorithm, respectively. The extracted features are then used to predict the cognitive assessment screening instrument (CASI) and mini mental state examination (MMSE) scores of the subjects using three different prediction models, namely a linear regression model, a nonlinear regression model, and a feedforward neural network (FNN) model. It is shown that the optimal prediction performance (i.e., prediction error = 7.89±5.86 for the CASI score and prediction error = 3.21±2.86 for the MMSE score) is obtained using the SFFS feature selection method and the FNN model. Overall, the results show that the feature selection and modeling methods proposed in this paper provide an accurate and objective means of evaluating the cognitive ability of individuals for dementia diagnosis purposes.
AB - An impaired cognitive ability is an important indicator of dementia-related disease. Accordingly, this paper utilizes 12 inertia-related features, 19 gait-related features, and 2 balance-related features to analyze the gait performance of subjects while performing single-task and dual-task walking tests and balance tests. The features most closely correlated with the cognitive ability of the subjects are extracted via a correlation analysis method and a sequential forward floating selection (SFFS) algorithm, respectively. The extracted features are then used to predict the cognitive assessment screening instrument (CASI) and mini mental state examination (MMSE) scores of the subjects using three different prediction models, namely a linear regression model, a nonlinear regression model, and a feedforward neural network (FNN) model. It is shown that the optimal prediction performance (i.e., prediction error = 7.89±5.86 for the CASI score and prediction error = 3.21±2.86 for the MMSE score) is obtained using the SFFS feature selection method and the FNN model. Overall, the results show that the feature selection and modeling methods proposed in this paper provide an accurate and objective means of evaluating the cognitive ability of individuals for dementia diagnosis purposes.
UR - http://www.scopus.com/inward/record.url?scp=85042718848&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85042718848&partnerID=8YFLogxK
U2 - 10.1109/JSEN.2018.2809478
DO - 10.1109/JSEN.2018.2809478
M3 - Article
AN - SCOPUS:85042718848
VL - 18
SP - 3338
EP - 3350
JO - IEEE Sensors Journal
JF - IEEE Sensors Journal
SN - 1530-437X
IS - 8
ER -