Limits...
An Ambulatory System for Gait Monitoring Based on Wireless Sensorized Insoles.

González I, Fontecha J, Hervás R, Bravo J - Sensors (Basel) (2015)

Bottom Line: The system can be used in gait analysis mobile applications, and it is designed for real-time demarcation of gait phases.Additionally, to provide a solution that is insensitive to perturbations caused by non-walking activities, a probabilistic classifier is employed to discriminate walking forward from other low-level activities, such as turning, walking backwards, lateral walking, etc.The combination of these two algorithms constitutes the first approach towards a continuous gait assessment system, by means of the avoidance of non-walking influences.

View Article: PubMed Central - PubMed

Affiliation: MAmI Research Lab, University of Castilla-La Mancha, Esc. Sup. de Informática, Paseo de la Universidad, 4, 13071 Ciudad Real, Spain. ivan.gdiaz@uclm.es.

ABSTRACT
A new gait phase detection system for continuous monitoring based on wireless sensorized insoles is presented. The system can be used in gait analysis mobile applications, and it is designed for real-time demarcation of gait phases. The system employs pressure sensors to assess the force exerted by each foot during walking. A fuzzy rule-based inference algorithm is implemented on a smartphone and used to detect each of the gait phases based on the sensor signals. Additionally, to provide a solution that is insensitive to perturbations caused by non-walking activities, a probabilistic classifier is employed to discriminate walking forward from other low-level activities, such as turning, walking backwards, lateral walking, etc. The combination of these two algorithms constitutes the first approach towards a continuous gait assessment system, by means of the avoidance of non-walking influences.

Show MeSH
GNB classifier's accuracy depending on the number of selected features from PCA.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4541895&req=5

f9-sensors-15-16589: GNB classifier's accuracy depending on the number of selected features from PCA.

Mentions: In Figure 9, the GNB classifier's accuracy during cross-validation analysis is plotted (y-axis) against the number of selected features (x-axis). The GNB classifier's accuracy follows a line of growth until a total of 22 features are selected, then holds a constant value around 90% to 92% accuracy and finally begins to decrease at 30 to 37 features (to 80% to 82% accuracy).


An Ambulatory System for Gait Monitoring Based on Wireless Sensorized Insoles.

González I, Fontecha J, Hervás R, Bravo J - Sensors (Basel) (2015)

GNB classifier's accuracy depending on the number of selected features from PCA.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4541895&req=5

f9-sensors-15-16589: GNB classifier's accuracy depending on the number of selected features from PCA.
Mentions: In Figure 9, the GNB classifier's accuracy during cross-validation analysis is plotted (y-axis) against the number of selected features (x-axis). The GNB classifier's accuracy follows a line of growth until a total of 22 features are selected, then holds a constant value around 90% to 92% accuracy and finally begins to decrease at 30 to 37 features (to 80% to 82% accuracy).

Bottom Line: The system can be used in gait analysis mobile applications, and it is designed for real-time demarcation of gait phases.Additionally, to provide a solution that is insensitive to perturbations caused by non-walking activities, a probabilistic classifier is employed to discriminate walking forward from other low-level activities, such as turning, walking backwards, lateral walking, etc.The combination of these two algorithms constitutes the first approach towards a continuous gait assessment system, by means of the avoidance of non-walking influences.

View Article: PubMed Central - PubMed

Affiliation: MAmI Research Lab, University of Castilla-La Mancha, Esc. Sup. de Informática, Paseo de la Universidad, 4, 13071 Ciudad Real, Spain. ivan.gdiaz@uclm.es.

ABSTRACT
A new gait phase detection system for continuous monitoring based on wireless sensorized insoles is presented. The system can be used in gait analysis mobile applications, and it is designed for real-time demarcation of gait phases. The system employs pressure sensors to assess the force exerted by each foot during walking. A fuzzy rule-based inference algorithm is implemented on a smartphone and used to detect each of the gait phases based on the sensor signals. Additionally, to provide a solution that is insensitive to perturbations caused by non-walking activities, a probabilistic classifier is employed to discriminate walking forward from other low-level activities, such as turning, walking backwards, lateral walking, etc. The combination of these two algorithms constitutes the first approach towards a continuous gait assessment system, by means of the avoidance of non-walking influences.

Show MeSH