Alzheimer disease prediction model based on decision fusion of CNN-BiLSTM deep neural networks

초록

Alzheimer’s disease (AD) is a chronic neurodegenerative disorder. Early prediction of Alzheimer’s progression is a crucial process for the patients and their families. As a chronic disease, AD data are multimodal and time series in nature. Building a deep learning model to optimize multi-objective cost function produces a more stable and accurate model. In this paper, we propose a multimodal multitask deep learning model for AD progression detection based five time series modalities and a collection of static data. The model predicts AD progression as a multi-class classification task and four critical cognitive scores as regression tasks. The experimental results show that our model is medically intuitive, more accurate, and more stable than the state-of-the-art studies. © Springer Nature Switzerland AG 2021.

키워드