Επιτομή:
With a focus on the ideas of machine learning and the use of the techniques offered, this thesis will introduce the basic concepts of machine learning and highlight the idea of edge computing. Everything that will be mentioned aims to form the framework for the creation and evaluation of a machine learning algorithm in order to be used in an application in the field of orthopedic rehabilitation. After becoming familiar with the use of the Edge Impulse platform and the Arduino Nano 33 BLE sense development board, an attempt is made to formulate a machine learning algorithm that will manage to evaluate and classify the flexion-extension movement of the human knee at a specific angle. The separation of the angles of movement is considered important as, based on medical advice, during orthopedic rehabilitation (from surgery), the flexibility of the leg in the specific angles also determines the success of the rehabilitation process. Through explaining the main points of the Edge Impulse platform, we come to 3 cases of algorithms that are evaluated in terms of the accuracy they provide in their predictions. To emphasize the importance of edge computing in resource and energy management, the most efficient algorithm is selected, developed on the Arduino Nano 33 BLE sense development board and with appropriate changes to the source code, the decision of the algorithm is sent to a mobile device, removing the need for another system for the operation of our application.
