Braak neurofibrillary stage specific ordinal regression analysis of miRNA expression data.

Ordinal logistic regression analysis (OLR) was implemented in order to detect miRNAs with monotonic expression patterns across Braak neurofibrillary stages. Braak stages were recorded during autopsy for each subject, and specific CSF subgroups consisted of stage 1 (n = 21), stage 2 (n = 21), stage 3 (n = 58), stage 4 (n = 37), stage 5 (n = 22), and stage 6 (n = 25) samples. For SER, Braak subcategories comprised stage 1 (n = 21), stage 2 (n = 27), stage 3 (n = 44), stage 4 (n = 31), stage 5 (n = 23), and stage 6 (n = 18) samples. Delta AIC () quantifies the information loss associated with using each model relative to the best approximating model. We report predictor variables with the lowest Akaike Information Criterion (AIC) and that satisfy assumptions of the OLR. p-Value* is unadjusted.为筛选出在Braak神经原纤维缠结分期中呈现单调表达模式的微小RNA（miRNAs），本研究采用有序逻辑回归分析（OLR）。所有受试者的Braak分期均经尸检确认，其对应的特异性脑脊液（CSF）亚组样本按分期划分为：1期（n = 21）、2期（n = 21）、3期（n = 58）、4期（n = 37）、5期（n = 22）及6期（n = 25）。针对SER队列，其Braak亚分类样本则为：1期（n = 21）、2期（n = 27）、3期（n = 44）、4期（n = 31）、5期（n = 23）及6期（n = 18）。ΔAIC用于量化相较于最优近似模型，使用各模型时所带来的信息损失量。本研究报告了赤池信息准则（AIC）值最低且符合有序逻辑回归分析假设的预测变量，带*号的p值为未校正结果。