A web-based framework using a Model-View-Controller architecture for human motion analysis

In a world with large technological advances and where technology takes a more and more essential place, the interaction between technology and health is increasingly present, particularly when one speaks of web technologies. The computerized clinical decision support systems (CDDS) are a good example of combination of web technology and health [1]. This work was part of a larger project that aimed to develop a tractable cloud-based open-source framework for human movement analysis and classification, using inverse kinematics from marker trajectories collected by means of a motion capture system. The main objective of the present work was to contribute to the development of a web-based framework for human motion analysis, throughout an interface designed under a Model-View-Controller (MVC) architecture [2]. The communication between the client and the server is based on the HyperText Transfer Protocol (HTTP) and the WebSocket standard. HTML5 is the base technology, which controls the overall layout of the front-end (user interface). Together with this technology we used Impress.js, which controls the style and layout of the web page. For a high-performance back-end, we used the Python programming language, which was responsible for the connection between the user interface and the tools of the OpenSim software (Scale and IK), as well as all the data processing. Usability and learnability assessment was assessed using the System Usability Scale [3]. The target audience for this study was chosen from two groups, which are representative of the potential end-user populations: a) Biomedical engineering students; b) Clinicians and students of physiotherapy. None of the groups had previous knowledge with the system, and before the experience, the individuals of the each groups had a 4 minutes explanation on the interface of the features and way of using. Preliminary results shown usability scores comprised between 74,2 and 84,4 respectively for groups a) and b), and learnability scores of 67,9 and 78,6 respectively for groups a) and b). With the proposed framework the users can simply upload the patient's clinical information, determine how the selected musculoskeletal model anthropometry should be modified so it best matches patients characteristics, and to what degree each model's segment (markers) should match the collected motion data during the inverse kinematics process. Finally a report is generated, where it is possible to observe a given variable, a group of variables compare, experimental and normative data, and also add comments, print, and save it in PDF. From SUS results we conclude that the interface was extremely useful, clear, and easy-to-use and learn for both groups. Also, the users are likely to recommend our framework to other.