Quantitative Assessment of Gait in Spinocerebellar Ataxia Using Deep Learning

Spinocerebellar ataxia is a neurological disorder characterized by impaired neural coordination, leading to unstable gait, coordination difficulties, and other motor abnormalities. This disease significantly impacts patients' daily lives, making it challenging for them to maintain balance, walk smoothly, and even perform fine motor skills. The causes of spinocerebellar ataxia are diverse, including genetic factors, brain injuries, or certain diseases. Many patients with spinocerebellar ataxia require specialized medical care and ongoing monitoring to manage its effects. This study aims to utilize Human Posture Estimation to detect the gait patterns of patients with spinocerebellar ataxia and integrate it with a deep learning model to distinguish between normal and affected individuals. The outcomes offer physicians an objective diagnostic tool and facilitate early detection and intervention for patients with impaired gait, thereby supporting the enhancement or preservation of their quality of life. The current prediction accuracy for the risk of ataxia gait is 80%, with an F1 score of 78.26%. It is necessary to conduct a detailed quantitative assessment by analyzing the gait of patients with spinocerebellar ataxia. Quantitative evaluation of the data allows for an objective assessment of the patient's gait, which helps healthcare professionals to more accurately understand and assess the gait status of patients with spinocerebellar ataxia.