Recent advances in extracting and processing rich semantics from medical texts

Abstract

Pharmaceuticalcompanies, healthcare organisations and individual patients exploiting advances in translational medicine and informational infrastructure, are now increasingly recording detailed patient records, an activity that was traditionally limited to only clinical interest. The traditional clinical records already comprised a broad range of clinical documents including nurse letters, discharge summaries and radiology reports describing a patient’s health status, diagnoses, applied procedures and observations of the health care team. The rich semantics such as facts, experiences, opinions or information that are hidden in those medical documents can now be combined with further information extracted from background documents such as clinical guidelines, documentation from medical trials and research literature, as well as self-reporting by patients. The semantics from these combined sources could – when extracted automatically – support a broad range of applications including clinical decision support systems, outcome analysis, cohort analysis, etc. Physicians could learn about the experiences of their colleagues, get hints to critical events in the treatment of a specific patient or receive information for improving treatments. Furthermore, such extracted semantics can be used to improve the general healthcare process by updating medical guidelines, identifying unexpected treatment interactions, monitoring healthcare quality, etc. Finally, there is an increased awareness that rich semantics such as sentiments, opinions and other qualitative factors are relevant in ensuring individualized care. Research on this topic has been presented in two workshops held in conjunction with the ESWC 2016 and AIME 2017 conferences. In this special issue we present extended versions of selected workshop papers, but also new research is included.