JMIR Publications

ABSTRACT

Background:

The exploration of unstructured data contained in the electronic health records (EHRs) holds the potential to positively impact clinical practice as well as research in Crohn’s disease (CD), an inflammatory bowel disease in which lesions may be found anywhere along the gastrointestinal tract. There is a clinical need for a rapid and early extraction of this information, for which the use of the most advanced technology is advantageous. In this regard, the EHRead® Technology is a natural language processing (NLP) system that was developed by SAVANA, to retrieve prominent biomedical information from narratives in the clinical notes contained in EHRs.

Objective:

The aim of this study was to validate the EHRead® Technology in identifying CD reports from clinical narratives.

Methods:

We employed the EHRead® Technology to explore and extract CD-related clinical information from EHRs. To validate this tool, we compared the EHRead® output with the gold standard (manually reviewed by experienced researchers) in order to assess the quality of NLP-identified records containing any reference to CD and its related variables.

Results:

The validation metrics for the main variable, (i.e. CD), were a precision of 0.88, a recall of 0.98 and an F1-score of 0.93. Regarding the secondary variables, we obtained a precision of 0.91, a recall of 0.71 and F1-score of 0.80 for CD disease flare, while the variable vedolizumab (treatment) yielded a precision, recall and F1-score of 0.86, 0.94 and 0.90, respectively.

Conclusions:

This evaluation demonstrates the ability of the EHRead® technology to identify CD patients and their related variables in unstructured information contained in EHRs. To the best of our knowledge, the present study is the first to use an NLP system for the identification of Crohn’s disease in reports written in Spanish.