Research And Application Of Electronic Medical Records Retrieval Based On The Correlation Algorithm Of Latent Semantic

With the rapid development of China’s medial insurance building and the deepening of the hospital information system development, electronic medical records as the basis of clinical information is becoming the core of the modern hospital information systems. Electronic medical records record the patient’s condition and other detailed and complete information. That can help the medical staff to better clinical decision-making. Today, electronic medical records increasingly widespread application in clinical diagnosis and treatment. At the same time, it generated a lot of electronic medical records data. It is a very important issue to effectively take advantages of these electronic medical records to help doctors to diagnose disease and medical research. The face of a large modern hospital electronic medical record data, it is a huge challenge for us to query the information fast and precisely that the medical staff needs. However, the information retrieval systems of electronic medical records based on traditional keywords and VSM have a certain amount of defects. For example, they can’t solve synonyms and polysemy of the medical words. Because of this kind of situation, it greatly affected the performance of electronic medical records retrieval. Therefore, it is particularly important to establish an intelligent electronic medial records retrieval system.In view of the polysemys and synonyms of the medical words, the paper analyzes the two kinds of intelligent information retrieval systems:latent semantic analysis and probabilistic latent semantic analysis. It is from the semantic level not the keywords level for them to realize information retrieval, and up from the concept of significance to understand and handle to retrieve the user’s request. This paper contributes on the following aspects:1. The author has built a more comprehensive the test set for electronic medical records retrieval.This test set is essential for evaluation of retrieval performance of electronic medical records.However, we don’t have a standard relevant test set of electronic medical records retrieval for our experiments in our country. So this paper has built the corpus. And experiments in this paper show that the corpus is built very reasonable, and the retrieval algorithms studied in the paper have also been a good evaluation.2. The latent semantic analysis algorithm is successfully applied to the field of electronic medical records. The author has realized the retrieval system of electronic medical records which is based on the latent semantic analysis.3. The probabilistic latent semantic analysis algorithm is successfully applied to the field of electronic medical records. The author has realized the retrieval system of electronic medical records which is based on the probabilistic latent semantic analysis.4. The author has designed an automatic algorithm for the selection of K value which is the number of the latent topics in PLSA algorithm. It is achieved by the cloer and closer to the current optimal value of K. And the paper uses this algorithm to replace the traditional exhaustive selection method. It is verified that the method in this paper which is for the selection of K value is better than the existing PLSA algorithm. And the K value selected which is the number of the latent topics is reasonable.5. The author put forward an algorithm based on combination of LSA and PLSA which is applied to the retrieval system of electronic medical records. The similarity of the combination algorithm is got by mixed calculating the similarity of the electronic medical records retrieval based on the latent semantic analysis and the electronic medical records retrieval based on probabilistic latent semantic analysis.And experimental results show that the proposed method can effectively improve the performance of the retrieval system for the electronic medical records.