Detection of Parkinson's cognitive impairment patients based on motor and non-motor symptoms

Gait measurement is an objective analysis method that can detect abnormal gait, facilitate early disease identification, and support doctors in formulating rehabilitation treatment plans. In recent years, cognitive scales and biomarkers have been widely used to detect Parkinson's cognitive impairment (PD-MCI). However, a lack of objective and reliable detection methods for PD-MCI exists. This study aims to identify patients with Parkinson's cognitive impairment utilizing two approaches: gait and executive function. Non-contact measurement was employed to obtain the gait signal, while the clock drawing test was utilized to obtain the executive function score of the patients. Ten PD-MCI patients and 10 ordinary PD were tested, and the kinematics data of the subjects during walking were recorded. Using patients' gait parameters and executive function scores as data characteristics, machine learning methods were utilized to classify and identify patients. The most suitable machine learning method for PD-MCI patient detection was discovered utilizing this approach. The results indicate that the XGBoost algorithm can identify and classify PD-MCI patients with more than 90% accuracy, providing effective support for doctors in identifying Parkinson's cognitive impairment early and customizing the most suitable rehabilitation treatment plan. These findings have significant research significance.