Extracting Daily Dosage from Medication Instructions in EHRs: An Automated Approach and Lessons Learned
This addresses the need for accurate daily dosage data to support medication timelines for physicians, though it is incremental as it builds on existing extraction methods.
The paper tackled the problem of extracting daily dosage from free-text medication instructions in EHRs by developing an automated approach combining deep learning, lexicons, and regex, achieving 0.98 precision and 0.95 recall on a dataset of 1,000 Sigs.
Medication timelines have been shown to be effective in helping physicians visualize complex patient medication information. A key feature in many such designs is a longitudinal representation of a medication's daily dosage and its changes over time. However, daily dosage as a discrete value is generally not provided and needs to be derived from free text instructions (Sig). Existing works in daily dosage extraction are narrow in scope, targeting dosage extraction for a single drug from clinical notes. Here, we present an automated approach to calculate daily dosage for all medications, combining deep learning-based named entity extractor with lexicon dictionaries and regular expressions, achieving 0.98 precision and 0.95 recall on an expert-generated dataset of 1,000 Sigs. We also analyze our expert-generated dataset, discuss the challenges in understanding the complex information contained in Sigs, and provide insights to guide future work in the general-purpose daily dosage calculation task.