Limits...
A wavelet-based approach to fall detection.

Palmerini L, BagalĂ  F, Zanetti A, Klenk J, Becker C, Cappello A - Sensors (Basel) (2015)

Bottom Line: Since recorded falls result in a non-stationary signal, a wavelet transform was chosen to examine fall patterns.The idea is to consider the average fall pattern as the "prototype fall".In order to detect falls, every acceleration signal can be compared to this prototype through wavelet analysis.This result suggests that the proposed wavelet-based feature is promising and future studies could use this feature (in combination with others considering different fall phases) in order to improve the performance of fall detection algorithms.

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi", University of Bologna, 40136 Bologna, Italy. luca.palmerini@unibo.it.

ABSTRACT
Falls among older people are a widely documented public health problem. Automatic fall detection has recently gained huge importance because it could allow for the immediate communication of falls to medical assistance. The aim of this work is to present a novel wavelet-based approach to fall detection, focusing on the impact phase and using a dataset of real-world falls. Since recorded falls result in a non-stationary signal, a wavelet transform was chosen to examine fall patterns. The idea is to consider the average fall pattern as the "prototype fall".In order to detect falls, every acceleration signal can be compared to this prototype through wavelet analysis. The similarity of the recorded signal with the prototype fall is a feature that can be used in order to determine the difference between falls and daily activities. The discriminative ability of this feature is evaluated on real-world data. It outperforms other features that are commonly used in fall detection studies, with an Area Under the Curve of 0.918. This result suggests that the proposed wavelet-based feature is promising and future studies could use this feature (in combination with others considering different fall phases) in order to improve the performance of fall detection algorithms.

Show MeSH
ROC curve of three features: the Wavelet-based (in blue), the Upper Peak Value (UPV, in red), and the Lower Peak Value (LPV, in green).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4482005&req=5

sensors-15-11575-f003: ROC curve of three features: the Wavelet-based (in blue), the Upper Peak Value (UPV, in red), and the Lower Peak Value (LPV, in green).

Mentions: As in [9], we considered sensitivity as the percentage of correctly detected falls, and specificity as the percentage of correctly detected ADLs. We obtained the ROC curve as follows: after the procedure explained in Section 2.2, all falls and all ADLs have a corresponding value of the wavelet-based feature. Generally, falls tend to have higher values (since they generally are more similar to the average fall pattern) and ADLs tend to have lower values. If we set a certain threshold, then all the recordings with a value over that threshold would be detected as falls while all the recordings with a value under that threshold would be detected as ADLs. By varying this threshold, we obtained different combinations of sensitivity and specificity which corresponded to different points in ROC curves in Figure 3.


A wavelet-based approach to fall detection.

Palmerini L, BagalĂ  F, Zanetti A, Klenk J, Becker C, Cappello A - Sensors (Basel) (2015)

ROC curve of three features: the Wavelet-based (in blue), the Upper Peak Value (UPV, in red), and the Lower Peak Value (LPV, in green).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4482005&req=5

sensors-15-11575-f003: ROC curve of three features: the Wavelet-based (in blue), the Upper Peak Value (UPV, in red), and the Lower Peak Value (LPV, in green).
Mentions: As in [9], we considered sensitivity as the percentage of correctly detected falls, and specificity as the percentage of correctly detected ADLs. We obtained the ROC curve as follows: after the procedure explained in Section 2.2, all falls and all ADLs have a corresponding value of the wavelet-based feature. Generally, falls tend to have higher values (since they generally are more similar to the average fall pattern) and ADLs tend to have lower values. If we set a certain threshold, then all the recordings with a value over that threshold would be detected as falls while all the recordings with a value under that threshold would be detected as ADLs. By varying this threshold, we obtained different combinations of sensitivity and specificity which corresponded to different points in ROC curves in Figure 3.

Bottom Line: Since recorded falls result in a non-stationary signal, a wavelet transform was chosen to examine fall patterns.The idea is to consider the average fall pattern as the "prototype fall".In order to detect falls, every acceleration signal can be compared to this prototype through wavelet analysis.This result suggests that the proposed wavelet-based feature is promising and future studies could use this feature (in combination with others considering different fall phases) in order to improve the performance of fall detection algorithms.

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi", University of Bologna, 40136 Bologna, Italy. luca.palmerini@unibo.it.

ABSTRACT
Falls among older people are a widely documented public health problem. Automatic fall detection has recently gained huge importance because it could allow for the immediate communication of falls to medical assistance. The aim of this work is to present a novel wavelet-based approach to fall detection, focusing on the impact phase and using a dataset of real-world falls. Since recorded falls result in a non-stationary signal, a wavelet transform was chosen to examine fall patterns. The idea is to consider the average fall pattern as the "prototype fall".In order to detect falls, every acceleration signal can be compared to this prototype through wavelet analysis. The similarity of the recorded signal with the prototype fall is a feature that can be used in order to determine the difference between falls and daily activities. The discriminative ability of this feature is evaluated on real-world data. It outperforms other features that are commonly used in fall detection studies, with an Area Under the Curve of 0.918. This result suggests that the proposed wavelet-based feature is promising and future studies could use this feature (in combination with others considering different fall phases) in order to improve the performance of fall detection algorithms.

Show MeSH