Avaliação quantitativa da rigidez de punho em pessoas com a doença de Parkinson por meio de uma órtese ativa

Abstract

Introduction and Literature Review: Rigidity is defined as an increase in joint resistance during passive movements, being an abnormality of muscle tone resulting from extrapyramidal disorders, with a constant degree of resistance regardless of the joint angle position. Although tremor is the most widely known motor signal, gait disturbances and the presence of rigidity in the upper and lower limbs are the most disabling for individuals with Parkinson's Disease (PD). Rigidity is one of the cardinal symptoms of PD and can be found in over 83% of individuals with PD. The method currently used by health professionals to measure rigidity is the Unified Parkinson’s Disease Rating System (UPDRS), a subjective method prone to evaluation errors. This thesis proposed, developed, and validated an objective method to quantify wrist rigidity using an instrumented orthosis. The objective was to differentiate individuals with PD from healthy controls (HC) and to correlate objective metrics with subjective clinical scores. Methodology: Methods used for the objective assessment of rigidity typically consist of inertial sensors to record movement, electromyographic sensors to register muscle activation, or servomotors; all these methods generally use flexion and extension movements of a joint to record signals. In this study, an instrumented wrist orthosis (force and current sensors) was used to apply passive flexion and extension movements to 30 volunteers: 15 with PD and 15 from the control group (HC). Twelve biomechanical features were extracted from the signals. Machine Learning (ML) models were trained to classify the groups (PD vs. HC). Statistical analyses, including Spearman’s correlation and multiple linear regression, were used to compare orthosis data with the MDS-UPDRS rigidity score. The device's usability was evaluated using the Short Flow State Scale (SFSS) and the System Usability Scale (SUS). Results: ML models demonstrated high classification capability, achieving 91.30% accuracy (QDA) with the full feature set and 86.96% (ET) after feature selection (CFS). A moderate and statistically significant correlation ρ=0,41 and p < 0,05 was found between the CARGA_Flexao (Flexion Load) feature and the MDS-UPDRS rigidity score. The regression model for rigidity quantification was able to explain 57.4% (Adjusted R²) of the clinical score variability, based on the interaction between biomechanical variables. The orthosis usability was rated as excellent, with global mean scores of 84.83 on the SUS scale (above the 68 threshold) and 4.20 on the SFSS scale, suggesting a state of flow. Conclusion: The results demonstrate that the instrumented orthosis is a viable, robust, and objective tool for assessing rigidity in PD. The instrumental quantification showed high classification accuracy and a significant association with clinical assessment. It is concluded that this system represents a promising technical subsidy to aid in the diagnosis and monitoring of PD, reducing the subjectivity of traditional evaluations.