Text mining for case report articles on "peritoneal dialysis" from PubMed database.

INTRODUCTION

The number of published medical articles on peritoneal dialysis (PD) has been increasing, and efficiently selecting information from numerous articles can be difficult. In this study, we examined whether artificial intelligence (AI) text mining can be a good support for efficiently collecting PD information.

METHODS

We performed text mining and analyzed all the abstracts of case reports on PD in the PubMed database. In total, 3137 case reports with abstracts related to "peritoneal dialysis" published from 1970 to 2021 were identified.

RESULTS

A total of 280 347 relevant words were extracted from all the abstracts. Word frequency analysis, word dependency analysis, and word frequency transition analysis showed that peritonitis, encapsulating peritoneal sclerosis, and child have been important keywords. Theseanalyses not only reflected historical background but also anticipated future trends of PD study.

CONCLUSION

These suggest that text mining can be a good support for efficiently collecting PD information.