Scalable Community Extraction of Text Networks for Automated Grouping in Medical Databases

Abstract

Networks are ubiquitous in today’s world. Community structure is a well-known feature of many empirical networks, and a lot of statistical methods have been developed for community detection. In this paper, we consider the problem of community extraction in text networks, which is greatly relevant in medical errors and patient safety databases. We adapt a well-known community extraction method to develop a scalable algorithm for extracting groups of similar documents in large text databases. The application of our method on a real-world patient safety report system demonstrates that the groups generated from community extraction are much more accurate than manual tagging by frontline workers.

Automated Error Labeling in Radiation Oncology via Statistical Natural Language Processing

“…One of the challenges of addressing medical errors is the manner in which errors are reported and the sheer amount of information provided in such reports [ 3 ]. Medical errors have traditionally been reported as free-text descriptions by front-line staff [ 4 ]. Inconsistencies occur as reporters use different levels of detail and various types of terminology and vocabulary for both technical and generic uses of language.…”

“…Furthermore, one of the primary barriers to error reporting is the varying perspective on errors themselves, since reporters may have conflicting perspectives regarding the incident being reported [ 5 ]. Given these problems associated with the analysis of error reports, natural language processing (NLP), the branch of machine learning that deals with human language and speech, has emerged as a potential technology in patient safety research to improve error prevention through better detection, reporting, and analysis [ 4 , 6 , 7 , 8 ]. A significant benefit of NLP models for medical personnel is the reduction in time and effort required to report medical errors.…”