Deep Convolutional Neural Network based Medical Concept Normalization

Medical concept normalization is a critical problem in information retrieval and clinical applications. In this paper, we focus on normalizing diagnostic and procedure names in Chinese discharge summaries to standard entities, which is formulated as a semantic matching problem. However, non-standard Chinese expressions, short-text normalization, heterogeneity of tasks and dynamic input of disambiguation entities pose critical challenges. Here we propose a basic model and a flexible model. The basic model solves ambiguous entities normalization. The flexible model deals with multiple dynamic input of entities. Specifically, in the basic model, we present a framework to disambiguate a disease and its corresponding operation simultaneously, which introduces a tensor generator and a novel multi-view convolutional neural network (CNN) with a multi-task shared structure. Multi-view CNN is adopted to synthesize them from different views. Then, the multi-task shared structure allows the model to exploit medical correlations among entities to better perform disambiguation tasks. Subsequently, we design a flexible model based on the basic model. Specifically, we add a flexible attention layer to all procedure vectors, and then apply a flexible multi-task scheme to share the correlated information. Comprehensive experiments indicate that our model outperforms existing baselines, demonstrating the effectiveness of our model.