DECAF: An interpretable deep cascading framework for ICU mortality prediction.

Medical risk detection is an important topic and a challenging task to improve the performance of clinical practices in Intensive Care Units (ICU). Although many bio-statistical learning and deep learning approaches have provided patient-specific mortality predictions, these existing methods lack interpretability that is crucial to gain adequate insight on why such predictions would work. In this paper, we introduce cascading theory to model the physiological domino effect and provide a novel approach to dynamically simulate the deterioration of patients' conditions. We propose a general DEep CAscading Framework (DECAF) to predict the potential risks of all physiological functions at each clinical stage. Compared with other feature-based and/or score-based models, our approach has a range of desirable properties, such as being interpretable, applicable with multi prediction tasks, and learnable from medical common sense and/or clinical experience knowledge. Experiments on a medical dataset (MIMIC-III) of 21,828 ICU patients show that DECAF reaches up to 89.30 % on AUROC, which surpasses the best competing methods for mortality prediction.

Gated Tree-based Graph Attention Network (GTGAT) for medical knowledge graph reasoning.

Knowledge graph (KG) is a multi-relational data that has proven valuable for many tasks including decision making and semantic search. In this paper, we present GTGAT (Gated Tree-based Graph Attention), a method for tackling the problems of transductive and inductive reasoning in generalized KGs. Based on recent advancement of graph attention network (GAT), we develop a gated tree-based method to distill valuable information in neighborhood via hierarchical-aware and semantic-aware attention mechanism. Our approach not only addresses several key challenges of GAT but is also capable of undertaking multiple downstream tasks. Experimental results have revealed that our proposed GTGAT has matched state-of-the-art approaches across transductive benchmarks on the Cora, Citeseer, and electronic medical record networks (EMRNet). Meanwhile, the inductive experiments on medical knowledge graphs show that GTGAT surpasses the best competing methods for personalized disease diagnosis.