| KEYWORD |
Design and validation of a machine learning-based approach for the identification of curved trajectories during walking in healthy and pathological conditions
keywords GAIT, KINEMATICS, MACHINE LEARNING, NEUROENGINEERING, SIGNAL PROCESSING
Reference persons VALENTINA AGOSTINI, MARCO GHISLIERI
Research Groups Biolab: Ingegneria Biomedica
Description Gait analysis often focuses on the study of human locomotion along rectilinear tracks. However, in the last decade, literature raised attention to the analysis of curvilinear walking, with applications in both rehabilitation engineering and pedestrian mobility. Indeed, in many cases, subjects navigate following curvilinear trajectories, rather than rectilinear ones, both in outdoor and indoor spaces. Recent works highlighted the importance of analyzing curved trajectories to investigate turning impairments in patients affected by Parkinson’s disease. In particular, when performing instrumented gait analysis along a hallway, it may be important to automatically segment epochs of rectilinear trajectories from curved trajectories. Expert operators often manually perform the selection of the rectilinear and curved trajectories by synchronously looking at gait data and video recordings. However, this procedure is time-consuming and highly affected by intra- and inter-operator variability. Alternative approaches, such as machine learning-based approaches, are being explored to perform signal classification and pattern recognition. This thesis aims at assessing the applicability of a novel machine learning-based approach for curved trajectory identification, specifically developed to overcome the main limitations of the standard approaches. The proposed approach will be also validated considering gait data acquired from subjects with different musculoskeletal or neurological disorders (e.g., Parkinson’s disease patients) to assess its applicability also in different pathological conditions.
Required skills Basic knowledge of MATLAB (data management and visualization), Signal processing, Basic knowledge of ML model training, and Neuroengineering exam passed with a score equal to or higher than 27/30.
Notes The proposal will be described during a meeting on Tuesday, July 18, 2023, at 11:00. VC Link: https://didattica.polito.it/pls/portal30/sviluppo.bbb_corsi.waitRoom?id=3614&p_tipo=DOCENTE
Deadline 25/07/2023
PROPONI LA TUA CANDIDATURA