Application Research Of Latent Semantic Analysis Applied In Clinical Record Text

With the rapid development of medical informatization, medical data is accumulating rapidly. Analyzing these data could provide great help to medical practitioners as well as patients. Performing quantitative analysis on Medical text, which is one of the loosest parts among all medical data, has always been a research focus duo to its complexity.Latent Semantic Analysis has been developing for years and is attracting ever-growing attention. It's applied in many fields including search indexing and recommendation.When applied to text analysis, Latent Semantic Analysis can discover hidden topics of the text without aid from linguistics. So it is suitable for handling loose and irregular text like clinical records.This paper applies Latent Semantic Analysis to clinical record analysis, and improves term weighting based on characteristics of sample records to get better effect. Tests show that the proposed method has improved performance on cardiology clinical records. This paper also researches on complication prediction using Latent Semantic Analysis. Proposed method can generate prediction for every unique cardiology clinical record.A distributed medical record retrieval service has been established. This paper tries to apply Latent Semantic Analysis to clinical records searching, provide both the traditional keyword-based method as well as a record-based searching method, which eliminates manual tagging and manual keywords extraction work.