Self-Care Problems Classification of Children with Physical and Motor Disability by Deep Neural Networks

Physical and motor disability is a disorder that greatly limits some of the individual main life activities. These disorders affect children in many countries of the world. In addition, it is a difficult process for physically and motorly disabled individuals to be classified by doctors with appropriate occupational treatments. Because, there are many variables that must be considered. The aim of this study is to classify the self-care skill problems of children with physical and motor disabilities by the minimal error using deep neural networks (DNN). For this purpose, DNN models with different parameters were created. The number of hidden layers, the number of neurons in the hidden layers, the activation function, the optimization algorithm, the loss function and the epoch value are taken into consideration in the creation of the models. The DSA models were trained and tested with the SCADI (Self-Care Activities Dataset based on ICFCY) data set. The classification performance of the models was demonstrated by using the F-1 score, precision (P), recall (R) and accuracy (ACC) metrics. Details of the 8 models with the best grading performance are presented. According to the findings, the best classification performance was obtained in the DSA-1 model using Adadelta optimization algorithm, Elu activation function and Categorical crossentropy loss function. The P, R, ACC and F1 scores of this model are 1. In other words, this model predicts the self-care skills problems of physical and motor disability children with 100% accuracy. In addition, in order to increase the validity of the three best models (DSA-1, DSA-2 and DSA-3), the training and testing process was performed with 10-fold cross-validation method. Mean cross validation accuracy values were calculated as 85.71%, 85.71% and 87.14% respectively. Occupational therapists can be used developed DSA models as a validating tool for diagnosing self-care problems