Tracking medication information across medical records

Area of Science:

• Medical Informatics
• Clinical Data Analysis
• Computational Medicine

Background:

• Electronic medical records (EMRs) contain extensive patient data, posing challenges for physicians to synthesize comprehensive medical histories.
• Efficiently tracking medication history, including drug names, dosages, and usage status, is crucial for patient care, especially for elderly or chronically ill individuals.

Purpose of the Study:

• To develop and evaluate a system for automatically tracking patient medications and their dosages over time using electronic medical records.
• To create a user-friendly timeline display of medication usage for improved clinical decision-making.

Main Methods:

• Utilized 923 reports from 11 patients at a university hospital.
• Identified drug names via dictionary look-up and dosages using regular expressions.
• Employed a support vector machine with expert-defined features to determine drug states (ON/OFF).

Main Results:

• Achieved promising results with an approximate recall of 87%.
• Successfully generated a timeline display illustrating patient medication history.
• Demonstrated the feasibility of automated medication tracking from EMRs.

Conclusions:

• The developed system shows significant potential for improving the review of patient medical histories by providing an automated medication timeline.
• This approach can alleviate the burden on physicians managing complex patient records.
• Further development is ongoing to refine the accuracy and scope of medication tracking.