Early Detection of Alzheimer's Disease with Blood Plasma Proteins Using Support Vector Machines

The successful development of amyloid-based biomarkers and tests for Alzheimer's disease (AD) represents an important milestone in AD diagnosis. However, two major limitations remain. Amyloid-based diagnostic provide limited information about the process they are unable to identify individuals with before significant amyloid-beta accumulation brain develops. objective this study is develop a method potential blood-based non-amyloid early detection. use blood attractive because it accessible relatively inexpensive. Our mainly based on machine learning (ML) techniques (support vector machines particular) their ability create multivariable models by patterns from complex data. Using novel feature selection evaluation modalities, we identified 5 panels proteins serve as AD. In particular, found that combination A2M, ApoE, BNP, Eot3, RAGE SGOT may be key biomarker profile disease. Disease detection achieved sensitivity (SN) > 80%, specificity (SP) 70%, area under receiver operating curve (AUC) at least 0.80 prodromal stage (with higher performance later stages) Existing ML performed poorly comparison disease, suggesting underlying protein not suitable results demonstrate feasibility using biomarkers.