Extended Study of Gait Analysis in Identification of Cardiovascular Diseases
Abstract
Walking pattern analysis known as Gait Analysis, is one of the key indicators among various clinical parameters such as cardiovascular disease, in identifying symptoms of diseases. With sensors, the walking patterns of people can be retrieved to identify the differences between patients and healthy controls. The paper comprises an analysis of Gait datasets comprising nine Kinematic gait parameters, using classification models, the development of a data augmentation algorithm, and a proposed prediction model. Classification analysis of the Gait dataset was done using Neural Network (98.65% accuracy). An algorithm GDAA was developed for the augmentation of Gait data whose time complexity is O(nf). The final result lies within a minimum, maximum range of the original dataset. The algorithm can be extended by researchers in augmenting their dataset with slight modifications. Data analysis of the augmented dataset was done on varying sizes of datasets to evaluate optimum classification results. Rigorous analysis of augmented data was done with Neural Network (97.1% accuracy). Ranking of gait features for both independent and dependent features was also done during the analysis. A prediction model is proposed which identifies whether input Gait data belongs to pathological gait or healthy gait with the help of a classification model trained on augmented data.