In recent years, machine learning, represented by deep learning, has become more and more in the field of health care. According to the type of data processed can be divided into numerical, textual and image data; This paper focuses on text data.
Clinical Diagnostic Decisions:
(Miotto r,et al;2016)  A new unsupervised depth feature learning method, which is a three-tiered, noise-cancelling automatic decoder, can be used to obtain a general patient characterization from the electronic health record data, which makes clinical predictive modeling more convenient. Used to capture the hierarchical regularity (hierarchical regularities) and dependencies of 700,000 patient aggregation electronic health records in Mt. Sinai Data Warehouse. The results were significantly better than other research methods using data representation based on the original electronic health files, as well as other feature learning strategies. "Deep patients" are far ahead in predicting the severity of diabetes, schizophrenia and various cancers. Findings suggest that the use of deep learning in electronic health records can be characterized by patient characterization, which can help us improve clinical prognosis, and also provide a deep learning framework for enhancing clinical decision-making systems. Similarly (Nguyen p,et al;2016)  Build a "depth record" (DEEPPR) based on a deep convolutional network to improve clinical diagnosis. (Nie L, Wang M, Zhang l, et al;2015)  A sparse deep learning framework was proposed to establish a user-based presentation of health characterization information to infer possible diseases.
Knowledge Base Building
(Savova G K, Masanz J J, Ogren P V, et al;2010) provides a comprehensive description of how to extract knowledge from clinical texts in architecture design, modules, and evaluation applications, which are mainly focused on word segmentation in the clinical field of natural language processing, Named body recognition and shallow level syntax parsing. (De Bruijn B,et al; 2010)  in the field of health, the identification of the entity name and the relationship analysis has given a benchmark effect. Mainly through the construction of natural language features and external knowledge features and auxiliary to SVM,CRF and other machine learning models. (Lei J, Tang B, Lu X, et al,2014) measure the effect of the CRF,SVM,ME,SSVM model on the recognition of the named body in the Chinese diagnostic text, and the results of the SSVM are 93.51% and 90.01% of the F-value in the Admission summary text and the discharge diagnostic sheet. With the deep application of deep learning in natural language (Collobert R, Weston J, Bottou L, et al,2011), some traditional natural language tasks have also gained some degree of improvement from the feature generation mechanism of deep learning. For example (Wu Y, Jiang M, Lei J, et al;2015) find the task of naming body recognition for traditional medicine, with unsupervised word vectors as input layer, the establishment of deep neural network framework can overcome the traditional CRF model.
miotto R, Li L, Kidd B A, et al deep patient:an unsupervised representation to Predict the future of Patients from th E Electronic Health Records[j]. Scientific Reports, 2016, 6.
nguyen P, Tran T, Wickramasinghe N, et al deepr:a convolutional Net for Medical Records[j]. ArXiv preprint arxiv:1607.07519, 2016.
nie L, Wang M, Zhang l, et al disease inference from health-related questions via sparse deep learning[j]. IEEE transactions on Knowledge and Data Engineering, 2015, 27 (8): 2107-2119.
savova G K, Masanz J J, Ogren P V, et al. Mayo Clinical Text Analysis and Knowledge Extraction System (ctakes): Archite Cture, component evaluation and APPLICATIONS[J]. Journal of the American Medical Informatics Association, 2010, 17 (5): 507-513.
uzunerö, South B R, Shen S, et al. I2b2/va Challenge on concepts, assertions, and relations in clinical text[j]. Journal of the American Medical Informatics Association, 2011, 18 (5): 552-556.
lei J, Tang B, Lu X, et al. A comprehensive study of named entity recognition in Chinese clinical text[j]. Journal of the American Medical Informatics Association, 2014, 21 (5): 808-814.
wu Y, Jiang M, Lei J, et al Named entity recognition in Chinese clinical text using deep neural network[j]. Studies in health technology and Informatics, 2015, 216:624.
collobert R, Weston J, Bottou L, et al Natural language Processing (almost) from scratch[j]. Journal of machine Learning (in): 2493-2537.