Feature Selection and Predictors of Falls with Foot Force Sensors Using KNN-Based Algorithms

Abstract

The aging process may lead to the degradation of lower extremity function in the elderly population, which can restrict their daily quality of life and gradually increase the fall risk. We aimed to determine whether objective measures of physical function could predict subsequent falls. Ground reaction force (GRF) data, which was quantified by sample entropy, was collected by foot force sensors. Thirty eight subjects (23 fallers and 15 non-fallers) participated in functional movement tests, including walking and sit-to-stand (STS). A feature selection algorithm was used to select relevant features to classify the elderly into two groups: at risk and not at risk of falling down, for three KNN-based classifiers: local mean-based k-nearest neighbor (LMKNN), pseudo nearest neighbor (PNN), local mean pseudo nearest neighbor (LMPNN) classification. We compared classification performances, and achieved the best results with LMPNN, with sensitivity, specificity and accuracy all 100%. Moreover, a subset of GRFs was significantly different between the two groups via Wilcoxon rank sum test, which is compatible with the classification results. This method could potentially be used by non-experts to monitor balance and the risk of falling down in the elderly population.

Faller Classification in Older Adults Using Wearable Sensors Based on Turn and Straight-Walking Accelerometer-Based Features

“…Normalization of features allows faster model training [38,45]. Each feature value in a participant’s feature set was normalized to the range [0, 1] as follows: $y_{normalized}=\frac{y-y_{min}}{y_{max}-y_{min}},$ where y is a feature value from one participant, and y min and y max are the minimum and maximum values of that feature, respectively, across all participants within a training set for each cross-validation fold.…”

PAPPI: Personalized analysis of plantar pressure images using statistical modelling and parametric mapping

“…More recently, machine learning algorithms have been employed in order to classify an individual's plantar pressure measurement into patient or healthy control groups [25][26][27][28][29]. In these studies, a database of plantar pressure data is combined with the corresponding group memberships in order to define a non-linear regression function between the two quantities.…”

“…In these studies, a database of plantar pressure data is combined with the corresponding group memberships in order to define a non-linear regression function between the two quantities. A variety of machine learning algorithms have been used to perform this regression, from artificial neural networks [25,27], to logistic regression [29], to nearest neighbour classification [28], to support vector machines [26]. Regardless of the algorithm used, the resulting classifier produces a personalized result: an individual's plantar pressure measurement, as a whole, gets labelled as either healthy or unhealthy.…”

Feature selection for the classification of fall-risk in older subjects: a combinational approach using static force-plate measures

“…Recent studies on fall-risk predictors have recognised the necessity for multiple features and non-linear algorithms in classifying fall-risk by exploring the potential of machine-learning to create effective classification models (25)(26)(27)(28)(29). The majority of these studies have focused on dynamic movement analysis to predict elderly fall-risk (25,26), while the presence of a thorough feature selection process has been lacking with one study of note using principal component analysis (30). The aim of this study is to demonstrate an effective feature selection procedure that can allow for more accurate and reliable classification of fall-risk among older subjects using static force-platform measures only.…”