Algorithms for automated sensorized L-test. Development and testing on seven patient populations

In recent decades, the study of human movement analysis has revealed itself to be increasingly important, as it appears to be directly related to people's psychophysical well-being. Its absence or deficiency causes and is caused by innumerable diseases. quantitative clinical tests, thanks to measuring instruments, obtain new measurement parameters in motion analysis. A new test, recently created for amputees, is called the Functional Mobility L-test. It's similar to the Timed Up and Go, but with 2 extra turns of 90 degrees each. The purpose of this thesis is to use this new test in other populations and above all to make it sensorized in order to extract new quantitative features. The test was applied to 7 different populations, Parkinson's disease, multiple sclerosis, people with lower limb amputation, patients with a proximal femur fracture, people with chronic heart failure, chronic obstructive pulmonary disease, and healthy adults. Each of them wore a wearable sensor on their lower back to subdivide the test in order to extract specific features. Postural transitions, such as getting up from a chair and sitting down, and the turns performed within the test, were identified through a series of algorithms implemented based on the TUG literature. By subsequently validating the algorithms, with a manual segmentation, the results obtained were very good for the turn algorithm, especially for the population with PD and PFF, and good for that of the stand and sit. The next step in the future will be to validate the algorithms using a gold standard, such as stereophotogrammetry, and include more populations with motor impairment.