Using Natural Language Processing to Identify Different Lens Pathology in Electronic Health Records

Using Electronic Health Record Data to Determine the Safety of Aqueous Humor Liquid Biopsies for Molecular Analyses

“…A recent study found that only about 50% of cases with lens pathology could be detected by analyzing International Classification of Diseases billing codes, whereas a free-text search-based algorithm identified >95% of the cases. 15 Analyzing surgical complications at large scale requires tedious manual review of operative and postoperative notes, especially since different surgeons may use different language to describe intraoperative events. The requirement to manually read surgical reports can significantly limit the number of cases that can be reviewed.…”

“…It has previously been shown that electronic health record (EHR) models that combine structured data with free-text searches demonstrate improved accuracy. 15 , 16 , 17 Here, we used established EHR software tools and applied a free-text searching strategy to assess complications in surgical reports and postoperative clinical notes. This allowed us to analyze the safety profile of AH biopsies at large scale, involving 1418 cases that were performed during intraocular surgery by many surgeons in a variety of cases.…”