Background: With the accelerating aging process, the number of participants with Alzheimer's disease (AD) is rising sharply, causing a huge economic burden.
Objective: This study aimed to identify blood protein and metabolic biomarkers and explore the diagnostic model for AD among elderly in southeast China.
Methods: We established a cohort among population with high risk AD in Zhejiang Province in 2018. Case and control groups each consisting of 45 subjects, matched for gender and age, were randomly selected from the cohort. Based on bioinformatics research, PRM/MRM technology was used to detect candidate biomarkers. Ensemble-based feature selection and machine learning methods was used to screen important variables as risk indicators for AD. Based on the risk biomarkers, the risk diagnostic model of AD in the elderly was constructed and evaluated.
Results: Cystine and CPB2 were evaluated as biomarkers. The diagnostic model is constructed using logistic regression algorithm with the best cutoff value, sensitivity, specificity, and accuracy of 0.554, 0.895, 0.976, and 0.938, respectively, which determined by Youden's index. The results showed that the model with protein and metabolite had a high efficiency.
Conclusion: It showed that the diagnostic model constructed by Cystine and CPB2 had a good performance on sample classification. This study was of great significance for the early screening and diagnosis of AD, timely intervention, control and delay the development of dementia in southeast China.
Keywords: Alzheimer’s disease; biomarker; diagnostic model; elderly.