Text Mining Approach for Categorizing Medication Errors

What is it? What does it do?

Medication errors are events that lead to the mistaken use of medication by a patient or provider that can cause harm to a patient. While medication errors are one of the most frequently reported patient safety report types and represent high potential harm to a patient, they are also a mostly preventable event. The FDA states that they receive more than 100,000 medication error reports annually.

Patient safety event (PSE) reporting systems in hospitals have the potential to dramatically improve the safety and quality of care by exposing possible vulnerabilities, like medication error, before they lead to an adverse event. However, due to the large volume of reports and perceived level of patient harm, many safety reports go unanalyzed.

Using natural language processing (NLP), these novel machine learning algorithms can help uncover themes in PSE reports using a semi-automatic approach. By categorizing medication-related reports into eight modified National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) classifications by programmatically processing PSE reports, these algorithms bring coherent themes in medication error reports to light. This approach can improve the categorization of medication-related PSE reports, which may lead to better identification of important medication safety hazards.