Final published version
Licence: CC BY: Creative Commons Attribution 4.0 International License
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Privacy-Preserving Wandering Behavior Sensing in Dementia Patients Using Modified Logistic and Dynamic Newton Leipnik Maps
AU - Shah, S.Y.
AU - Ahmad, J.
AU - Masood, F.
AU - Shah, S.Y.
AU - Pervaiz, H.
AU - Taylor, W.
AU - Imran, M.A.
AU - Abbasi, Q.H.
PY - 2021/2/1
Y1 - 2021/2/1
N2 - The health status of an elderly person can be identified by examining the additive effects of aging along disease linked to it and can lead to the 'unstable incapacity'. This health status is essentially determined by the apparent decline of independence in Activities of Daily Living (ADLs). Detecting ADLs provide possibilities of improving the home life of elderly people as it can be applied to fall detection systems.. This article looks at Radar images to detect large scale body movements. Using a publicly available Radar spectogram dataset, Deep Learning and Machine Learning techniques are used for image classification of Walking, Sitting, Standing, Picking up Object, Drinking Water and Falling Radar spectograms. The Machine Learning algorithm used were Random Forest, K Nearest Neighbours and Support Vector Machine. The Deep Learning algorithms used in this article were Long Short Term Memory, Bi-directional Long Short-Term Memory and Convolutional Neural Network. In addition to using Machine Learning and Deep Learning on the spectograms, data processing techniques such as Principal Component Analysis and Data Augmentation is applied to the spectogram images. The work done in this article is divided into 4 experiments. The first experiment applies Machine and Deep Learning to the the Raw images data, the second experiment applies Principal Component Analysis to the Raw image Data, the third experiment applies Data Augmentation to the Raw image data and the fourth and final experiment applies Principal Component Analysis and Data Augmentation to the Raw image data. The results obtained in these experiments found that the best results were obtained using the CNN algorithm with Principal Component Analysis and Data Augmentation together to obtain a result of 95.30 % accuracy. Results also showed how Principal Component Analysis was most beneficial when the training data was expanded by augmentation of the available data.
AB - The health status of an elderly person can be identified by examining the additive effects of aging along disease linked to it and can lead to the 'unstable incapacity'. This health status is essentially determined by the apparent decline of independence in Activities of Daily Living (ADLs). Detecting ADLs provide possibilities of improving the home life of elderly people as it can be applied to fall detection systems.. This article looks at Radar images to detect large scale body movements. Using a publicly available Radar spectogram dataset, Deep Learning and Machine Learning techniques are used for image classification of Walking, Sitting, Standing, Picking up Object, Drinking Water and Falling Radar spectograms. The Machine Learning algorithm used were Random Forest, K Nearest Neighbours and Support Vector Machine. The Deep Learning algorithms used in this article were Long Short Term Memory, Bi-directional Long Short-Term Memory and Convolutional Neural Network. In addition to using Machine Learning and Deep Learning on the spectograms, data processing techniques such as Principal Component Analysis and Data Augmentation is applied to the spectogram images. The work done in this article is divided into 4 experiments. The first experiment applies Machine and Deep Learning to the the Raw images data, the second experiment applies Principal Component Analysis to the Raw image Data, the third experiment applies Data Augmentation to the Raw image data and the fourth and final experiment applies Principal Component Analysis and Data Augmentation to the Raw image data. The results obtained in these experiments found that the best results were obtained using the CNN algorithm with Principal Component Analysis and Data Augmentation together to obtain a result of 95.30 % accuracy. Results also showed how Principal Component Analysis was most beneficial when the training data was expanded by augmentation of the available data.
KW - human activity
KW - machine learning
KW - patient monitoring
KW - Wandering behavior
KW - wireless sensing
KW - Brain
KW - Classification (of information)
KW - Convolutional neural networks
KW - Decision trees
KW - Deep learning
KW - Image analysis
KW - Learning systems
KW - Long short-term memory
KW - Nearest neighbor search
KW - Potable water
KW - Privacy by design
KW - Radar
KW - Support vector machines
KW - Activities of daily living (ADLs)
KW - Additive effects
KW - Data augmentation
KW - Data processing techniques
KW - Dementia patients
KW - K-nearest neighbours
KW - Machine learning techniques
KW - Privacy preserving
KW - Learning algorithms
U2 - 10.1109/JSEN.2020.3022564
DO - 10.1109/JSEN.2020.3022564
M3 - Journal article
VL - 21
SP - 3669
EP - 3679
JO - IEEE Sensors Journal
JF - IEEE Sensors Journal
SN - 1530-437X
IS - 3
ER -