Parkinson's disease (PD) is a neurodegenerative disorder in which dopaminergic medications, such as levodopa, are typically used to improve motor symptoms and the overall level of people's mobility. To enhance and personalise clinical management, it is important to assess the adherence and impact of pharmacological treatments on motor states (such as ON/OFF/DISKYNESIA, to cite a few). In this context, in addition to clinical assessments performed by PD specialists, it becomes crucial to leverage digital health technologies (e.g., wearable devices) that can collect motor symptoms objectively, continuously and remotely, so as to monitor participants even in an uncontrolled environment. This work aims to implement and validate an automatic motor state identification algorithm based on a novel energy-based composite index capturing mobility and motor symptom fluctuations occurring during the day. This work aims to identify and validate an energy-based composite index, whose evaluation and comparison with a suitable computed threshold can help identifying different motor states and correlate the quantitative information with well-validated clinical scales, to give a double response: i) the efficacy of the pharmacological treatment during daily life (OFF states identification); ii) provide an accurate estimation of the clinical scale index, by computing its value through a function modelling based on energy computation. Preliminary results show that, as regards i), the OFF states were identified with an accuracy of around 98%, while ii) results are presented in terms of clustering robustness (silhouette coefficient of 0.82) and a paired t-test confirming the relation between quantitative computed index and clinical scores. These promising outcomes are opening the way for the development of algorithms able to analyse real-time data from wearable devices to assess disease progression in people with PD.
A Novel Energy-Based Composite Index for Assessing Motor State in Parkinson's Disease by Means of IMU-Based Digital Health Technology
Carissimo, C.Membro del Collaboration Group
;Cerro, G.Membro del Collaboration Group
;
2024-01-01
Abstract
Parkinson's disease (PD) is a neurodegenerative disorder in which dopaminergic medications, such as levodopa, are typically used to improve motor symptoms and the overall level of people's mobility. To enhance and personalise clinical management, it is important to assess the adherence and impact of pharmacological treatments on motor states (such as ON/OFF/DISKYNESIA, to cite a few). In this context, in addition to clinical assessments performed by PD specialists, it becomes crucial to leverage digital health technologies (e.g., wearable devices) that can collect motor symptoms objectively, continuously and remotely, so as to monitor participants even in an uncontrolled environment. This work aims to implement and validate an automatic motor state identification algorithm based on a novel energy-based composite index capturing mobility and motor symptom fluctuations occurring during the day. This work aims to identify and validate an energy-based composite index, whose evaluation and comparison with a suitable computed threshold can help identifying different motor states and correlate the quantitative information with well-validated clinical scales, to give a double response: i) the efficacy of the pharmacological treatment during daily life (OFF states identification); ii) provide an accurate estimation of the clinical scale index, by computing its value through a function modelling based on energy computation. Preliminary results show that, as regards i), the OFF states were identified with an accuracy of around 98%, while ii) results are presented in terms of clustering robustness (silhouette coefficient of 0.82) and a paired t-test confirming the relation between quantitative computed index and clinical scores. These promising outcomes are opening the way for the development of algorithms able to analyse real-time data from wearable devices to assess disease progression in people with PD.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
