Metodologia para avaliação objetiva da Bradicinesia na doença de parkinson

• Ano de defesa: 2022
• Autor(a) principal: Luiz, Luiza Maire David
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Engenharia Elétrica
• Palavras-chave em Português: Bradicinesia; Doença de Parkinson; Avaliação remota; Componentes da bradicinesia; Bradykinesia; Parkinson’s disease; Remote evaluation; Bradykinesia components; CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA; Engenharia Elétrica; Engenharia biomédica; Cérebro - Doença; Mal de parkinson
• Link de acesso: https://repositorio.ufu.br/handle/123456789/35527; http://doi.org/10.14393/ufu.te.2022.344

Parkinson's disease (PD) is a potentially, devastating and multifactorial neurodegenerative disorder, linked to genetic and environmental factors, which compromises motor and non-motor functions. One difficulty in evaluating the PD progress is related to the prolonged time for clinical assessments and questionnaires application, the complexity and heterogeneity of motor impairments and the need for a specialized analysis. Bradykinesia is a Parkinson's disease symptoms, characterized as the slowness in the execution of movements, such as the reaction time to start the movements, the increase the range of movement during the execution of the tasks time for the duration of the motor pattern, the rapid fatigue in performing prolonged tasks and changing rhythm during activities. The assessment of bradykinesia severity can be influenced by the examiner's time of experience, by the patient's cooperation, as by the variability of inter-examiner results. The Movement Disorder Society (MDS) highlighted the need to develop and validate technologies for remote assessment of motor condition in PD. In this sense, this research presents two distinct strategies for the bradykinesia assessment, which are: (i) intra and inter-rater remote assessment of bradykinesia in Parkinson's disease and (ii) objective assessment, using inertial sensors, with the aim of discriminating the bradykinesia components: slowness, amplitude and rhythm. To the first objective (i), a mediation team performed and recorded the clinical evaluation MDS-UPDRS (Movement Disorder Society – Unified Parkinson’s Disease Rating Scale) and edited all the videos, to later available the videos to 14 evaluators with varying experience, between 3 and 10 years, in Parkinson's disease evaluation. For intra and inter-rater analysis, statistical methods were applied between evaluations, Cohen's Kappa, intraclass coefficient correlation (ICC) and the Bland-Altman method. The correlation results obtained between experience raters are strong. Difficulty in clinical assessments can be influenced by lack of training and difficulty using clinical scales. Correlation matches also revealed low agreement due inexperience and poorly trained examiners. Despite flaws in clinical assessments, a large number of clinical raters, using blinded methods and gold standard rating scales, provide a more robust “fundamental truth” of bradykinesia assessment. It is important to highlight that the entire study execution was only possible due the mediator group compliance with the protocol. The effectiveness and feasibility of telemedicine during the COVID-19 pandemic is considered a breakthrough in medical care, enabling remote access, even outside the pandemic context. In the second objective of thesis (ii), the aim was to evaluate the bradykinesia components: slowness, amplitude and rhythm, extracted from inertial signals. The experimental protocol consisted of four tasks perform: finger tapping, open and close the hand, pronation and supination and flexion and extension. For that, inertial sensors were positioned at the upper limb for movements analysis. Fifteen individuals with Parkinson's disease and fifteen healthy individuals with matched sex and age performed this protocol. Signal processing was performed in R software and three feature extraction methods were used. The algorithm SFA (Slow Feature Analysis) was applied to extract the slowness characteristic. An index was created to extract the amplitude, based on the total amplitude and interquartile amplitude, and the rhythm was quantified by the variation coefficient. The component indices were correlated to the clinical assessment and statistical metrics for inter-group comparison. The obtained results indicate strong correlations between the bradykinesia components and clinical assessments, between -0.9 and -0.8 for amplitude and slowness, respectively. And for intergroup comparisons, the three indices showed a statistically significant difference. With this study, it can be concluded that the processing tools chosen to extract the characteristics of the inertial sensors were adequate, innovated and efficient in the discrimination of the components of bradykinesia.