Assessment of motor function in Parkinson disease: study and development of algorithms and mobile solutions

The measure of the mobility in the subjects suffering from Parkinson's disease is currently an important research topic because, in such complex pathology, movement alteration represents one of the most debilitating phenomena that patients have to face. This is accompanied by the need to monitor the evolution of the disease from a quantitative point of view, with the use of an instrumented methodology developed directly for this purpose. The detailed information about the evolution of the pathology allows the doctor to plan treatments tailoring them to the patient's needs. In addition to the characterization of the disease, identifying the first signs at an early stage of the pathology is of main importance: as a result clinicians are seeking help in new technologies. The use of wearable sensors for the analysis of movement occupies a large part of the research in this field since the beginning of our century. In the last decade, the diffusion in clinics of wearable systems is exponentially growing because they are perceived as a way to monitor the patient in the hospital and/or at home, and because, wearable sensors are a more comfortable and quick way of carrying out quantitative analyses, compared to widely-used methods such as optoelectronic systems. Moreover, the increasing distribution of smartphones has made it possible for the smartphone to interface with the wearable sensors to make them even more appealing and eye-catching to the user. The purpose of this research is the use of wearable inertial sensors, connected with Bluetooth to a smartphone that synchronize and save the acquired data in order to: on the one hand discriminate the presence or not of the pathology in subjects that are located at an early stage of the disease, and in the other hand, to look for correlations of results with what is commonly used by neurologists at evaluation, i.e. UPDRS and H&Y evaluation scales. The hardware involved in tests was customized from existing devices and custom made Android applications for data acquisition and synchronization were designed and developed during this study. The whole thesis work has been organized around a physical test that is common in clinical practice , namely the Timed Up and Go Test, that can be instrumented for its objective evaluation by means of wearable sensing nodes. TUG test instrumentation took the start from a redundant experimental setup, involving the use of many magneto inertial sensing units positioned on the shanks, wrists, chest and body center of mass (BCOM). Participants were asked to walk for a 3-meters TUG test as it is common for the compilation of the UPDRS but they also performed an extended version of the test, the 7-meters TUG, to obtain a greater number of step during the gait phase to get higher reliability in gait parameter extraction. Data analysis show that the BCOM-placed sensor is the one that allows extracting more information when used alone. In fact using BCOM sensor data, it was possible to discriminate not only the postural transition and the turning phase during the test with Sensitivity > 74.7 and Specificity > 78.2 for both PD and CTRL group, but also to extract temporal and spatial parameters such as step time, stride time, cadence and stride length without showing significant differences (p-value > α, α=0.05) compared to those obtained from right and left shanks sensors. Among them, the step length calculated in the PD group of women has proved to be important in the discrimination of the presence or not of the disease. The stride time estimated in the same group showed a significant correlation with the H&Y and UPDRS evaluation scale (R^{2}>0.7). Moreover, the evaluation of the trunk tilt in the women with the device secured on the chest showed a correlation with the UPDRS and H&Y evaluation scale with values of R^2>0.6 . Study results allowed to simplify the experimental setup to a single-unit based configuration. BCOM sensor in fact allowed to extract both parameters which can discriminate against the presence or not of the pathology, and calculate parameters that show correlation with the evaluation scales that represent the gold standard from a medical point of view.These results pave the way to new scenarios for patient monitoring, that could also be based on the use of the inertial sensors embedded on the smartphone. In fact, such a solution would allow to monitor movement alteration during specific tasks and to track his/her movement quality in everyday life. This would allow a reduction of psychological load due to the presence of the medical staff and then obtain a more detailed and truthfully follow up assessment that could be a daily one instead of being performed on a six-months basis.