Simulation of MobiFall Dataset for Fall Detection Using MATLAB Classifier Algorithms

Fall accidents are considered one of the significant global public health concerns, and the largest proportion of fatal accidents are experienced by elderly people. Currently, there is a demand for creating an effective machine learning-based fall detection system that is significantly portable at a low cost. Public datasets are available for simulating an effective classifier for development. Hence, the current study is aimed at simulating the MobiFall fall detection dataset to acquire an effective machine learning classifier. The methodology included a study of the various fall detection systems as well as general features for machine learning classifications. The most suitable potential combination of machine learning algorithms that will provide the best accuracy, precision, sensitivity, specificity and lowest training time was developed via simulation models using MATLAB. Input data was selected from the MobiFall dataset for simulations. Up to 23 algorithms, including Decision Trees, Discriminant Analysis, Naive Bayes Classifiers, Support Vector Machines, KNN and available Ensemble Classifiers, were simulated. Four sets of experiments were done using accelerometers with varying features and cross-validation. Currently, a combination of Quadratic SVM, Cubic SVM and Fine KNN was chosen to be used as the most appropriate classifier to train a fall detection system.