The Proactive Rehabilitation Sensing System (PReSenS) is the component in charge of exercise definition, monitoring and evaluation. The healthcare expert and the patient front-ends have been built using a Model-View-Controller software design. We have used Microsoft's Kinect as the selected depth sensor to build a working prototype able to acquire, recognize and evaluate rehabilitation exercise performances. The healthcare expert application is able to create exercises, save their templates from the sensor to the database, and design exercise plans. In addition, it includes a motion analysis module to analyze the motion data graphically, from both the template and patient recordings, and to export these data to external formats. On the other hand, the patient front-end is able to show an exercise plan for a patient, giving the opportunity to choose an exercise of the plan and make a recording of the patient performance. This recording is carried out unobtrusively since the application saves the quaternion information of the patient joints for each recorded frame, for its later evaluation and analysis by the system. Finally, the Web front-end contains a system for the management of patient and exercise plans, and also a 3D viewer that makes possible to visualize the recorded exercises using a Web browser together with the evaluation of the patient performance by healthcare experts.

The Exercise Editor, in the previous image, allows the medical expert to record template exercises and to define which joints are relevant for their performance. The 3-D view eases the visualization of motion by means of virtual camera rotations and translations. We can also track the quality of the sensor signal through the Skeleton Visualization panel, to ensure that the template is correctly saved and stored into the database.

Once a template performance of an exercise has been acquired from the depth sensor, the Recording Editor might be used to edit the recorded data, as an easy-to-use and simple video editor, to select and cut those frames of relevance for the exercise and save them as template. The following image shows an example for the edition of the recording template of exercise Flamenco, which is composed by a single pose:

On the other hand, the Patient Front-End of PReSenS allows both medical experts and patients to manage exercise plans daily and weekly. The medical experts proposes a plan to be performed by patients, and they can visualize their daily and weekly plans, select exercises to be performed, and carry out the plan exercises. After an exercise performance has been finished by a patient, the system provides an automatic score about the performance and saves all the data to the database, so that the medical expert is able to visualize it and make finer evaluations and feedback to the patient. The following figures shows an example of the Patient's Plan Window for a sample patient Allen Key, and the automatic evaluation after performing an exercise of his plan:

Finally, PReSenS also includes a simple tool for time series representation of exercise performance, both for template and user recording. With this tool, medical experts can make an in-depth analysis of the movement of patients during the exercise performance, including several options for parameter selection and visualization. The following two figures show an example of the time series representation tool. The first one plots the evolution in time of the angles between the left and right shoulders in respect to the chest, and the left and right elbows in respect to the shoulders, for the exercise Both Hands Up. The second one describes the relation of the angle between the left and right elbows in respect to the left and right shoulders and how they evolve in time, for the same exercise.