Abstract: “Clinical notes in medical institutions are essential not only for information sharing among medical staff but also\ for analysis to improve medical activities. Especially, the record of free description is important for grasping the situation of the patients. However, manually recording has the problem of input labor and fluctuation of expression by the person who records. In order to solve this problem and enable statistical analysis and mechanical analysis, templates are expected. This paper proposes an interactive support system for defining medical record template for each disease by extracting feature words corresponding to the disease from medical records using machine learning. We constructed an interactive system that displays the context of those characteristic words as a directed graph. A pull-down template can be generated from this graph. We developed a system to generate such graphs interactively for 123,736 free description clinical records of a hospital and examined its”

Keywords: Interactive, Visualization, Template Graph, Clinical Notes