This work presents a non-invasive low-cost system suitable for the at home assessment of the neurological impairment of patients affected by Parkinson's Disease. The assessment is automatic and it is based on the accurate tracking of hands and fingers movements of the patient during the execution of standard upper limb tasks specified by the Unified Parkinson's Disease Rating Scale (UPDRS). The system is based on a human computer interface made by light gloves and an optical tracking RGB-Depth device. The accurate tracking and characterization of hands and fingers movements allows both the automatic and objective assessment of UPDRS tasks and the gesture-based management of the system, making it suitable for motor impaired users as are PD patients. The assessment of UPDRS tasks is performed by a machine learning approach which use the kinematic parameters that characterize the patient movements as input to trained classifiers to rate the UPDRS scores of the performance. The classifiers have been trained by an experimental campaign where cohorts of PD patients were assessed both by a neurologist and the system. Results on the assessment accuracy of the system, as compared to neurologist's assessments, are given along with preliminary results on monitoring experiments at home.
Steps toward Automatic Assessment of Parkinson's Disease at Home
R Nerino;C Ferraris;G Pettiti;A Chimienti;
2018
Abstract
This work presents a non-invasive low-cost system suitable for the at home assessment of the neurological impairment of patients affected by Parkinson's Disease. The assessment is automatic and it is based on the accurate tracking of hands and fingers movements of the patient during the execution of standard upper limb tasks specified by the Unified Parkinson's Disease Rating Scale (UPDRS). The system is based on a human computer interface made by light gloves and an optical tracking RGB-Depth device. The accurate tracking and characterization of hands and fingers movements allows both the automatic and objective assessment of UPDRS tasks and the gesture-based management of the system, making it suitable for motor impaired users as are PD patients. The assessment of UPDRS tasks is performed by a machine learning approach which use the kinematic parameters that characterize the patient movements as input to trained classifiers to rate the UPDRS scores of the performance. The classifiers have been trained by an experimental campaign where cohorts of PD patients were assessed both by a neurologist and the system. Results on the assessment accuracy of the system, as compared to neurologist's assessments, are given along with preliminary results on monitoring experiments at home.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.