Abstract:Background: Mobility and balance is essential for older adults' well-being and independence and the ability to maintain physically active. Early identification of functional impairment may enable early risk-of-fall assessments and preventive measures. There is a need to find new solutions to assess functional ability in easy, efficient, and accurate ways, which can be clinically used frequently and repetitively. Therefore, we need to understand how functional tests and expert assessments (EAs) correlate with n… Show more
“…A large dataset will open up the possibility of testing additional classifiers and retesting the classifiers that are already tested and mentioned in this paper. Commonly used deep ML methods on skeletal data such as Convolutional Neural Network (CNN) [ 44 ] and Recurrent Neural Network (RNN) [ 45 ] will also be attempted in order to verify their usability with the proposed and additional set of skeletal parameters for skill assessment. Skeletal features that were not considered in this paper will also be tested for usability in these classifiers.…”
Section: Discussion and Future Directionmentioning
Early and self-identification of locomotive degradation facilitates us with awareness and motivation to prevent further deterioration. We propose the usage of nine squat and four one-leg standing exercise features as input parameters to Machine Learning (ML) classifiers in order to perform lower limb skill assessment. The significance of this approach is that it does not demand manpower and infrastructure, unlike traditional methods. We base the output layer of the classifiers on the Short Test Battery Locomotive Syndrome (STBLS) test used to detect Locomotive Syndrome (LS) approved by the Japanese Orthopedic Association (JOA). We obtained three assessment scores by using this test, namely sit-stand, 2-stride, and Geriatric Locomotive Function Scale (GLFS-25). We tested two ML methods, namely an Artificial Neural Network (ANN) comprised of two hidden layers with six nodes per layer configured with Rectified-Linear-Unit (ReLU) activation function and a Random Forest (RF) regressor with number of estimators varied from 5 to 100. We could predict the stand-up and 2-stride scores of the STBLS test with correlation of 0.59 and 0.76 between the real and predicted data, respectively, by using the ANN. The best accuracies (R-squared values) obtained through the RF regressor were 0.86, 0.79, and 0.73 for stand-up, 2-stride, and GLFS-25 scores, respectively.
“…A large dataset will open up the possibility of testing additional classifiers and retesting the classifiers that are already tested and mentioned in this paper. Commonly used deep ML methods on skeletal data such as Convolutional Neural Network (CNN) [ 44 ] and Recurrent Neural Network (RNN) [ 45 ] will also be attempted in order to verify their usability with the proposed and additional set of skeletal parameters for skill assessment. Skeletal features that were not considered in this paper will also be tested for usability in these classifiers.…”
Section: Discussion and Future Directionmentioning
Early and self-identification of locomotive degradation facilitates us with awareness and motivation to prevent further deterioration. We propose the usage of nine squat and four one-leg standing exercise features as input parameters to Machine Learning (ML) classifiers in order to perform lower limb skill assessment. The significance of this approach is that it does not demand manpower and infrastructure, unlike traditional methods. We base the output layer of the classifiers on the Short Test Battery Locomotive Syndrome (STBLS) test used to detect Locomotive Syndrome (LS) approved by the Japanese Orthopedic Association (JOA). We obtained three assessment scores by using this test, namely sit-stand, 2-stride, and Geriatric Locomotive Function Scale (GLFS-25). We tested two ML methods, namely an Artificial Neural Network (ANN) comprised of two hidden layers with six nodes per layer configured with Rectified-Linear-Unit (ReLU) activation function and a Random Forest (RF) regressor with number of estimators varied from 5 to 100. We could predict the stand-up and 2-stride scores of the STBLS test with correlation of 0.59 and 0.76 between the real and predicted data, respectively, by using the ANN. The best accuracies (R-squared values) obtained through the RF regressor were 0.86, 0.79, and 0.73 for stand-up, 2-stride, and GLFS-25 scores, respectively.
“…In this study, we have used assessment and measurement results collected from our previous studies 1,6 as shown on Figure 1. It includes: three supervised FTs (TUG, 30sCST, and 4SBT) performed in a controlled environment, EA of the sit-to-stand movement performed on video recordings by an experienced physiotherapist using the Instrument for movement analysis of person transfer and mobility (IRAF) in daily living, 8 daily life PA collected by an ActivPAL device within 7 days, and completed SA questionnaire about daily life PA.…”
Section: Methodsmentioning
confidence: 99%
“…In order to compare the 2D SAT with the 3D SAT approaches, we applied the same number of experiments (8 settings: indirect/direct features, cut/uncut sequences, normalized/not normalized features) with the same deep-learning models: convolutional neural network (CNN) and recurrent neural network (RNN) as in our previous studies, 1,6 and shown on Figure 2.Figure 2.Machine learning experiments overview.…”
Section: Methodsmentioning
confidence: 99%
“…The FTs procedure usually involves trained personal (healthcare professionals) to assist with the test. In addition, an EA method 1 has been developed for assessing qualitative aspects of the movement performance, in which the movement performance is assessed by an assessor in the form of a score or a rate. SA methods are performed by individuals (older adults) through questionnaires, surveys, or interviews and are used for assessing mobility and balance status and measuring daily life PA. With AC methods, individuals should wear an accelerometer device for a period of at least seven days in order to measure daily life PA. As result, the accelerometer raw data is collected and a daily life physical activity data report is provided (e.g., hours of sitting, walking, biking, sit-to-stand transition, etc.).…”
Section: Introductionmentioning
confidence: 99%
“…In 3D SAT, a 3D sensor camera (Kinect) records human movements, estimates the joint position in each frame of the motion recording, and maps this information to a movement quality score with help of AI techniques. This tool was tested and validated for the assessment of mobility and balance among healthy older adults, 1 and for measuring PA in healthy older adults. 6 Both studies show that the 3D SAT is a good tool for predicting the results of the FTs and accelerometer-based assessed PA.…”
Background: Maintaining physical activity (PA) and functioning (mobility, balance) is essential for older adults’ well-being and quality of life. However, current methods (functional tests, self-reports) and available techniques (accelerometers, sensors, advanced movement analysis systems) for assessing physical activity and functioning have shown to be less reliable, time- and resource-consuming with limited routine usage in clinical practice. There is a need to simplify the assessment of physical activity and functioning among older adults both in health care and clinical studies. This work presents a study on using Skeleton Avatar Technology (SAT) for this assessment. SAT analyzes human movement videos using artificial intelligence (AI). The study compares handy SAT based on 2D camera technology (2D SAT) with previously studied 3D SAT for assessing physical activity and functioning in older adults. Objective: To explore whether 2D SAT yields accurate results in physical activity and functioning assessment in healthy older adults, statistically compared to the accuracy of 3D SAT. Method: The mobile pose estimation model provided by Tensorflow was used to extract 2D skeletons from the video recordings of functional test movements. Deep neural networks were used to predict the outcomes of functional tests (FT), expert-based movement quality assessment (EA), accelerometer-based assessments (AC), and self-assessments of PA (SA). To compare the accuracy with 3D SAT models, statistical analysis was used to test whether the difference in the predictions between 2D and 3D models is significant or not. Results: Overall, the accuracy of 2D SAT is lower than 3D SAT in predicting FTs and EA. 2D SAT was able to predict AC with 7% Mean Absolute Error (MAE), and self-assessed PA (SA) with 16% MAE. On average MAE was 4% higher for 2D than for 3D SAT. There was no significant difference found between the 2D and the 3D model for AC and for two FTs (30 seconds chair stand test, 30sCST and Timed up and go, TUG). A significant difference was found for the 2D- and 3D-model of another FT (4-stage balance test, 4SBT). Conclusion: Altogether, the results show that handy 2D SAT might be used for assessing physical activity in older adults without a significant loss of accuracy compared to time-consuming standard tests and to bulky 3D SAT-based assessments. However, the accuracy of 2D SAT in assessing physical functioning should be improved. Taken together, this study shows promising results to use 2D SAT for assessing physical activity in healthy older adults in future clinical studies and clinical practice.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
