Presenter Information

Benjamin BrewerFollow
Jay Dorris

Presentation Type

Poster Presentation

Abstract

This study evaluates the feasibility of using natural language processing (NLP) tools to identify errors in titratable infusion orders for Joint Commission readiness. Titratable infusions are adjusted at the bedside by nurses based on predefined parameters, such as monitoring goals, titration rate, conditions for adjustment, and maximum dose. These parameters are often documented as free text within order sets but can be inadvertently removed during the verification process. Some electronic health records (EHRs) may not prominently display this critical information, leading to potential Joint Commission violations. A synthetic dataset was generated using artificial intelligence to simulate one year’s worth of titratable infusion data from a large academic medical center. A Python-based formula was developed to detect errors and associate them with specific medications and patients. Data analysis was conducted in Google Colab. Validation was performed using four randomized datasets, each containing 400 synthetic orders. The formula was subsequently tested on a dataset of 25,000 orders. An expected error rate of 2% was built into the dataset, and upon evaluation, the formula correctly identified 500 erroneous orders. NLP tools demonstrate potential for enhancing Joint Commission preparedness by systematically identifying errors in titratable infusion orders. Their integration into EHR systems could improve compliance and patient safety.

Faculty Mentor

Dr. Jay Dorris

Share

COinS
 

Developing NLP-Based Open-Source Tools to Improve Titratable Infusion Order Interpretation in EHR Systems

This study evaluates the feasibility of using natural language processing (NLP) tools to identify errors in titratable infusion orders for Joint Commission readiness. Titratable infusions are adjusted at the bedside by nurses based on predefined parameters, such as monitoring goals, titration rate, conditions for adjustment, and maximum dose. These parameters are often documented as free text within order sets but can be inadvertently removed during the verification process. Some electronic health records (EHRs) may not prominently display this critical information, leading to potential Joint Commission violations. A synthetic dataset was generated using artificial intelligence to simulate one year’s worth of titratable infusion data from a large academic medical center. A Python-based formula was developed to detect errors and associate them with specific medications and patients. Data analysis was conducted in Google Colab. Validation was performed using four randomized datasets, each containing 400 synthetic orders. The formula was subsequently tested on a dataset of 25,000 orders. An expected error rate of 2% was built into the dataset, and upon evaluation, the formula correctly identified 500 erroneous orders. NLP tools demonstrate potential for enhancing Joint Commission preparedness by systematically identifying errors in titratable infusion orders. Their integration into EHR systems could improve compliance and patient safety.

 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.