Konwledge-Enabled Diagnosis Assistant Based on Obstetric EMRs and Knowledge Graph

Konwledge-Enabled Diagnosis Assistant Based on Obstetric EMRs and Knowledge Graph Kunli Zhang1,2, Xu Zhao1,2, Lei Zhuang1, Qi Xie1 and Hongying Zan1,2 1School of Information Engineering, Zhengzhou University, Zhengzhou, China 2Peng Cheng Laboratory, Shenzhen, China {ieklzhang,ielzhuang,ieqxie,iehyzan}@zzu.edu.cn zhaox917@163.com Abstract The obstetric Electronic Medical Record (EMR) contains a large amount of medical data and health information. It plays a vital role in improving the quality of the diagnosis assistant ser- vice. In this paper, we treat the diagnosis assistant as a multi-label classification task and propose a Knowledge-Enabled Diagnosis Assistant (KEDA) model for the obstetric diagnosis assistant. We utilize the numerical information in EMRs and the external knowledge from Chinese Ob- stetric Knowledge Graph (COKG) to enhance the text representation of EMRs. Specifically, the bidirectional maximum matching method and similarity-based approach are used to obtain the entities set contained in EMRs and linked to the COKG. The final knowledge representation is obtained by a weight-based disease prediction algorithm, and it is fused with the text represen- tation through a linear weighting method. Experiment results show that our approach can bring about +3.53 F1 score improvements upon the strong BERT baseline in the diagnosis assistant task. 1 Introduction Health service relations on the health of millions of people, and it is a livelihood issue in our coun- try. Specifically in China, which has a huge population, the total amount of medical resources is still insufficient. The imbalance between the supply and demand for medical services is still the focus of China’s healthcare industry. Although the implementation of China’s Universal Two-child Policy in 2016 achieved many benefits, it also leads to an increase in the proportion of older pregnant women and the incidence of various complications (Yang and Yang, 2016). Compared to the overall supply of the medical industry, the lack of obstetric medical resources is prominent. Since the issue of the Basic Norms of Electronic Medical Records (Trial) (China’s Ministry of Health, 2010) by the National Health and Family Planning Medical Affairs Commission in 2010, medical institu- tions have accumulated many obstetric Electronic Medical Records (EMRs). EMRs are detailed records of medical activities, dominated by the semi-structured or unstructured texts. There is a lot of medical knowledge and health information in EMRs, which is the core medical big data. The first course record in EMRs can be divided into the chief complaint, physical examination, auxiliary examination, admitting diagnosis, diagnostic basis, and treatment plan. In general, there is not a single diagnosis in the admit- ting diagnosis, it usually includes normal obstetric diagnosis, medical diagnosis, and complications. As a consequence, the diagnosis assistant task based on the Chinese obstetric EMRs can be treated as a multi- label text classification problem, in which the different diagnoses can be regarded as the variable labels. However, the doctor’s diagnosis and treatment process are based on comprehensive clinical experience and knowledge in the medical field to make a diagnosis and formulate a corresponding treatment plan. At the same time, they can also explain the corresponding diagnosis basis to the …

Content truncated to 500 words