Washington University in St. Louis, St. Louis, Missouri, USA.
University of Pittsburgh, Pittsburgh, Pennsylvania, USA.
Alzheimers Dement. 2021 Jun;17(6):1005-1016. doi: 10.1002/alz.12259. Epub 2021 Jan 21.
Machine learning models were used to discover novel disease trajectories for autosomal dominant Alzheimer's disease.
Longitudinal structural magnetic resonance imaging, amyloid positron emission tomography (PET), and fluorodeoxyglucose PET were acquired in 131 mutation carriers and 74 non-carriers from the Dominantly Inherited Alzheimer Network; the groups were matched for age, education, sex, and apolipoprotein ε4 (APOE ε4). A deep neural network was trained to predict disease progression for each modality. Relief algorithms identified the strongest predictors of mutation status.
The Relief algorithm identified the caudate, cingulate, and precuneus as the strongest predictors among all modalities. The model yielded accurate results for predicting future Pittsburgh compound B (R = 0.95), fluorodeoxyglucose (R = 0.93), and atrophy (R = 0.95) in mutation carriers compared to non-carriers.
Results suggest a sigmoidal trajectory for amyloid, a biphasic response for metabolism, and a gradual decrease in volume, with disease progression primarily in subcortical, middle frontal, and posterior parietal regions.
使用机器学习模型发现常染色体显性阿尔茨海默病的新疾病轨迹。
从显性遗传性阿尔茨海默病网络中采集了 131 名突变携带者和 74 名非携带者的纵向结构磁共振成像、淀粉样蛋白正电子发射断层扫描（PET）和氟脱氧葡萄糖 PET；两组在年龄、教育程度、性别和载脂蛋白 E4（APOE ε4）方面相匹配。训练深度神经网络以预测每种模态的疾病进展。Relief 算法确定了突变状态的最强预测因子。
Relief 算法确定尾状核、扣带回和楔前叶是所有模态中最强的预测因子。与非携带者相比，该模型在预测未来匹兹堡化合物 B（R = 0.95）、氟脱氧葡萄糖（R = 0.93）和萎缩（R = 0.95）方面，对突变携带者的预测结果非常准确。
结果表明淀粉样蛋白呈类正弦轨迹，代谢呈双相反应，体积逐渐下降，疾病进展主要发生在皮质下、中额和后顶叶区域。
