Research Outline
1. Research Question
Type 2 Diabetes Mellitus (T2DM) nearly doubles Alzheimer’s Disease (AD) risk, but plasma proteomics research on AD risk in T2DM populations and precise predictive tools for this group are lacking. The core question: How to screen key predictors via plasma proteomics and construct/validate an AD risk model for T2DM patients?
2. Research Objectives
(1) Based on UK Biobank (UKB) data, integrate plasma proteomics and clinical indicators, use machine learning to screen AD predictors in T2DM patients, and build AD risk model.
(2) Establish a longitudinal cohort of elderly T2DM patients at Zhujiang Hospital, conduct follow-up to verify the model’s generalization and robustness, and provide an early high-risk identification tool.
3. Scientific Basis
AD is a global neurodegenerative disease (60%-70% of dementia cases). The 2024 NIA-AA criteria prioritize biomarkers (e.g., plasma p-tau217), but existing biomarkers only reflect partial pathology, and CSF sampling is invasive. Plasma proteomics is promising: a study of over 3,300 individuals identified 416 AD-related plasma proteins, and a 7-protein model outperformed classic biomarkers like A!40/A!42.
T2DM and AD share brain insulin resistance as the core link. T2DM impairs the PI3K/AKT/mTOR pathway, increasing A! and tau hyperphosphorylation , and accelerates MCI-to-AD progression. However, plasma proteomics research on AD risk in T2DM patients is scarce, and existing biomarkers are less applicable here.
This study uses UKB’s large-sample data and Zhujiang Hospital’s resources. Combining plasma proteomics and machine learning, it fills gaps and provides a practical tool .