Author : Nayan Pawar, Dr. (Prof). Amol Potgantwar and Dr. (Prof).Mangesh Ghonge

Padua straight model is generally utilized for the hazard appraisal of venous thromboembolism (VTE), which is a typical and preventable confusion for inpatients. Separating VTE chance components from unstructured medicinal records in emergency clinic can comprehend VTE occasions and create productive hazard appraisal model. In this investigation, we proposed a philosophy based technique to mine VTE hazard factors joining natural language processing (NLP) and AI (ML) strategies. Medici- nal records of 3106 inpatients were prepared and terms in different ontologies from different segments of records improved in VTE patients were arranged consequently. At that point ML strategies were utilized to assess terms’ significance and terms in- side conceding analysis and progress notes indicated preferred VTE forecast execution over different segments. At last a novel VTE forecast model was assembled de- pendent on chose terms and demonstrated higher AUC score (0.815) than the Padua model (0.789).

Keywords: Medical Record, Venous thromboembolism (VTE), Natural Language Processing (NLP), Risk Assessment, Machine Learning (ML).