Based on the high-quality Parkinson's disease cohort population and biological sample database in the early stage of the team, multi-dimensional intelligent wearable device technology and machine learning were used to screen and classify digital biomarkers in the prodromal PD cohort, early PD cohort and healthy population cohort. Using peptide nanoprobes, metabolomics and Simoa technology to find the pathological molecular biomarkers of high-purity blood neurogenic exosomes in the prodromal stage of PD cohort, early PD cohort and healthy population cohort, and conduct classification and combination study. The association analysis was used to explore the specific internal relationship between digital biomarkers and neurogenic exosome molecular biomarkers, to explore the potential relationship between the two types of biomarkers, and to further apply machine learning methods to establish a fusion model for early diagnosis of PD including markers related to digital phenotype and pathological molecular mechanism, and to verify it clinically. The research results of this project are expected to bring new strategies for the early diagnosis of PD biomarkers, provide theoretical support for clinical transformation, and have important clinical significance for achieving the goal of early diagnosis and early treatment.