Identification of problems in picture archiving and communication systems management using text mining

ObjectiveThe purpose of this study was to identify the problems needing attention in PACS operation related to attributes of medical facilities.MethodsA questionnaire survey regarding PACS management was conducted in Hokkaido, in the northern part of Japan, and PACS related problems were described by radiologic technologists as in "additional comments" in the questionnaire. In order to identify the problems, we analyzed the comments using three methods: multidimensional scaling, cooccurrence networks, and self-organizing maps. Then, we used correspondence analysis (CA) in order to identify the relationships between the PACS problems and the characteristics of the medical facilities.ResultsThere were 93 responses, a response rate of 36%. The number of additional comments provided was 47 which represents about a half of all responses. We determined 14 types of problems. Of these fourteen, eight types with an appearance frequency of two or more were analyzed to identify the relationship between the problems and the characteristics of the facility by using correspondence analysis. The most frequent problem was "data transfer between facilities" and the second frequent problem was "system connection within a medical facility." "Network management," "training for system users" and "data transfer in system update" were characterized by "almost/clearly assigned" administrators. "Management of video data" were characterized by medium sized hospitals, and "nothing" was for smaller facilities.ConclusionsThis study identified 14 types of problems in dealing with PACS management and clarified the relationships between the PACS problems and the characteristics of the medical facilities.