Assessment of Balance Control Subsystems by Artificial Intelligence

Fall risk classification with posturographic parameters in community-dwelling older adults: a machine learning and explainable artificial intelligence approach

BACKGROUND

Computerized posturography obtained in standing conditions has been applied to classify fall risk for older adults or disease groups. Combining machine learning (ML) approaches is superior to traditional regression analysis for its ability to handle complex data regarding its characteristics of being high-dimensional, non-linear, and highly correlated. The study goal was to use ML algorithms to classify fall risks in community-dwelling older adults with the aid of an explainable artificial intelligence (XAI) approach to increase interpretability.

METHODS

A total of 215 participants were included for analysis. The input information included personal metrics and posturographic parameters obtained from a tracker-based posturography of four standing postures. Two classification criteria were used: with a previous history of falls and the timed-up-and-go (TUG) test. We used three meta-heuristic methods for feature selection to handle the large numbers of parameters and improve efficacy, and the SHapley Additive exPlanations (SHAP) method was used to display the weights of the selected features on the model.

RESULTS

The results showed that posturographic parameters could classify the participants with TUG scores higher or lower than 10 s but were less effective in classifying fall risk according to previous fall history. Feature selections improved the accuracy with the TUG as the classification label, and the Slime Mould Algorithm had the best performance (accuracy: 0.72 to 0.77, area under the curve: 0.80 to 0.90). In contrast, feature selection did not improve the model performance significantly with the previous fall history as a classification label. The SHAP values also helped to display the importance of different features in the model.

CONCLUSION

Posturographic parameters in standing can be used to classify fall risks with high accuracy based on the TUG scores in community-dwelling older adults. Using feature selection improves the model's performance. The results highlight the potential utility of ML algorithms and XAI to provide guidance for developing more robust and accurate fall classification models. Trial registration Not applicable.

Cutoff Point of Mini-Balance Evaluation Systems Test Scores for Elderly Estimated by Center of Pressure Measurements by Linear Regression and Decision Tree Classification

BACKGROUND

Understanding balance ability and assessing the risk of possible falls are very important for elderly rehabilitation. The Mini-Balanced Evaluation System Test (Mini-BESTest) is an important survey for older adults to evaluate subject balance, but it is not easy to complete due to various limitations of physical activities, including occasional fear of injury. A center of pressure (CoP) signal can be extracted from a force pressure plate with a short recording time, and it is relatively achievable to ask subjects to stand on a force pressure plate in a clinical environment. The goal of this study is to estimate the cutoff score of Mini-BESTest scores from CoP data.

METHODS

CoP signals from a human balance evaluation database with data from 75 people were used. Time domain, frequency domain, and nonlinear domain parameters of 60 s CoP signals were extracted to classify different cutoff point scores for both linear regression and a decision tree algorithm. Classification performances were evaluated by accuracy and area under a receiver operating characteristic curve.

RESULTS

The correlation coefficient between real and estimated Mini-BESTest scores by linear regression is 0.16. Instead of linear regression, binary classification accuracy above or below a cutoff point score was developed to examine the CoP classification performance for Mini-BESTest scores. The decision tree algorithm is superior to regression analysis among scores from 16 to 20. The highest area under the curve is 0.76 at a cutoff point score of 21 for the CoP measurement condition of eyes opened on the foam, and the corresponding classification accuracy is 76.15%.

CONCLUSIONS

CoP measurement is a potential tool to estimate corresponding balance and fall survey scores for elderly rehabilitation and is useful for clinical users.

Classification of Older and Fall-Experienced Subjects by Postural Sway Data using Mass Spring Damper Model

The quiet standing test is used to detect diseases of the postural control system. The descriptive statistics of the center of pressure (COP) of older people during the test tend to be larger than those of healthy young people, but they cannot indicate structural problems in postural control. COP trajectories can be mathematically modeled with structural parameters such as viscosity, stiffness, and stochastic terms; however, the classification accuracy of older and fall-experienced people using such parameters has not been sufficiently verified. In this study, six structural parameters of a mass-spring-damper (MSD) model were estimated using two datasets, in which a total of 212 subjects performed quiet standing tests under four conditions. The estimated parameters were used for classification with a random forest algorithm to examine the differences in classification accuracy compared to seven conventional descriptive statistics methods. For the classification of older subjects, the classification accuracy of the MSD parameter method was the highest in foam condition, with positive likelihood ratios approximately 8.0. For the classification of fall-experienced subjects, the positive likelihood ratio of the MSD parameter method was 5.0, which is better than conventional descriptive statistics. Various MSD parameters revealed that aging and changing the floor surface and visual conditions cause oscillations in the COP behavior. While the MSD parameters were confirmed to help classify older subjects more accurately than the conventional descriptive statistics, there was room for further improvement in the classification accuracy of fall-experienced subjects.

Deep Neural Networks for Human's Fall-risk Prediction using Force-Plate Time Series Signal

Early and accurate identification of the balance deficits could reduce falls, in particular for older adults, a prone population. Our work investigates deep neural networks' capacity to identify human balance patterns towards predicting fall-risk. Human balance ability can be characterized based on commonly-used balance metrics, such as those derived from the force-plate time series. We hypothesized that low, moderate, and high risk of falling can be characterized based on balance metrics, derived from the force-plate time series, in conjunction with deep learning algorithms. Further, we predicted that our proposed One-One-One Deep Neural Networks algorithm provides a considerable increase in performance compared to other algorithms. Here, an open source force-plate dataset, which quantified human balance from a wide demographic of human participants (163 females and males aged 18-86) for varied standing conditions (eyes-open firm surface, eyes-closed firm surface, eyes-open foam surface, eyes-closed foam surface) was used. Classification was based on one of the several indicators of fall-risk tied to the fear of falling: the clinically-used Falls Efficacy Scale (FES) assessment. For human fall-risk prediction, the deep learning architecture implemented comprised of: Recurrent Neural Network (RNN), Long-Short Time Memory (LSTM), One Dimensional Convolutional Neural Network (1D-CNN), and a proposed One-One-One Deep Neural Network. Results showed that our One-One-One Deep Neural Networks algorithm outperformed the other aforementioned algorithms and state-of-the-art models on the same dataset. With an accuracy, precision, and sensitivity of 99.9%, 100%, 100%, respectively at the 12th epoch, we found that our proposed One-One-One Deep Neural Network model is the most efficient neural network in predicting human's fall-risk (based on the FES measure) using the force-plate time series signal. This is a novel methodology for an accurate prediction of human risk of fall.

Review of the Upright Balance Assessment Based on the Force Plate

Quantitative assessment is crucial for the evaluation of human postural balance. The force plate system is the key quantitative balance assessment method. The purpose of this study is to review the important concepts in balance assessment and analyze the experimental conditions, parameter variables, and application scope based on force plate technology. As there is a wide range of balance assessment tests and a variety of commercial force plate systems to choose from, there is room for further improvement of the test details and evaluation variables of the balance assessment. The recommendations presented in this article are the foundation and key part of the postural balance assessment; these recommendations focus on the type of force plate, the subject's foot posture, and the choice of assessment variables, which further enriches the content of posturography. In order to promote a more reasonable balance assessment method based on force plates, further methodological research and a stronger consensus are still needed.