Assessment of Machine Learning Models for Classification of Movement Patterns During a Weight-Shifting Exergame
ABSTARCT :
In exercise gaming (exergaming), reward systems are typically based on rules/templates from joint movement patterns. These rules or templates need broad ranges in definitions of correct movement patterns to accommodate varying body shapes and sizes.This can lead to inaccurate rewards and, thus, inefficient exercise, which can be detrimental to progress.If exergames are to be used in serious settings like rehabilitation, accurate rewards for correctly performed movements are crucial.This article aims to investigate the level of accuracy machine learning/deep learning models can achieve in classification of correct repetitions naturally elicited from a weight-shifting exergame. Twelve healthy elderly (10F, age 70.4 SD 11.4) are recruited. Movements are captured using a marker-based 3-D motion-capture system.Random forest (RF), support vector machine, k-nearest neighbors, and multilayer perceptron (MLP) are the employed models, trained and tested on whole body movement patterns and on subsets of joints.MLP and RF reached the highest recall and F1-score, respectively, when using combined data from joint subsets. MLP recall range are 91% to 94%, and RF F1-score range 79% to 80%. MLP and RF also reached the highest recall and F1-score in each joint subset, respectively. Here, MLP ranged from 93% to 97% recall, while RF ranged from 73% to 80% F1-score. Recall results, show that >9 out of 10 repetitions are classified correctly, indicating that MLP/RF can be used to identify correctly performed repetitions of a weight-shifting exercise when using full-body data and when using joint subset data.
EXISTING SYSTEM :
This could provide a way of using ML in exergames to more accurately reward movements during play, thus ensuring movement quality to a greater extent than the existing systems do.Future work will focus on the use of ML models in actual exergame situations, as this possibly elicits movements that are noisier than in the current study, hence making the repetitions difficult to classify as being incorrect or correct.Using motion capture systems with lower accuracy, and only using e.g. shoulder JCPs as input to the classification models would also be interesting to test in an actual exergaming setting, to see if the movements are still different enough to be classified as being correctly or incorrectly performed with similar accuracy to this study
DISADVANTAGE :
? The high MLP performance is likely due to the manner MLP models adjust the weights and biases in an iterative manner for a given classification problem by using gradient descent .
? As such, MLP models also intuitively assess importance of different features during training.
? The model that performs best on average might not always be the best performing model in all problem subsets .
? This indicates that it is necessary to evaluate the specific problem at hand, and how different models perform with the given data types, available computational power and noise level.
PROPOSED SYSTEM :
? Collection of time series data was conducted November 2017 using a 10-camera, 100Hz, 3DMoCap system (Vicon Motion Systems Ltd). Simultaneous ground reaction force (GRF) data was collected using a 1000Hz force plate (Kistler Inc) embedded in the floor, and digital video in sagittal view was recorded for quality control purposes.
? Reflective markers were placed according to the Plug-in-Gait full-body biomechanical model, with head and hand markers excluded.
? This probably resulted in more variable JCP’s during correct repetitions, enabling the ML models to accurately identify them.
ADVANTAGE :
? In addition, the technologies that exergames are based on make it possible to tailor games to individual needs and goals , which is a major advantage that could make exergaming even more effective than traditional exercise.
? Furthermore, to ensure that exergames are appropriate for older adults, extensive research has been conducted into the design and usability requirements for this population, resulting in guidelines and design principles that apply to exergames for older adults
? The high MLP performance is likely due to the manner MLP models adjust the weights and biases in an iterative manner for a given classification problem by using gradient descent . As such, MLP models also intuitively assess importance of different features during training.
|