Creating a Synthetic Clinical Trial: Comparative Effectiveness Analyses Using an Electronic Medical Record

Area of Science:

• Biomedical Informatics
• Clinical Research Methodology
• Data Science in Healthcare

Background:

• Electronic medical records (EMRs) contain valuable data for clinical research.
• Extracting and analyzing EMR data is challenging due to technical complexities.
• A need exists for a simplified, portable system to leverage EMR data.

Purpose of the Study:

• To develop and validate a framework for overcoming EMR data extraction challenges.
• To create a simple, portable, and expandable system for utilizing EMR data.
• To generate synthetic clinical trials using EMR data.

Main Methods:

• Developed an Internet-accessible Python program to extract patient data based on user-defined criteria.
• Created a modular Worker Application to identify functional status, smoking status, and treatment classification.
• Validated the framework by creating a synthetic clinical trial mirroring a prospective randomized phase III clinical trial.

Main Results:

• Identified and extracted data for 281 lung cancer patients, matching inclusion criteria for a synthetic trial.
• Achieved high precision and accuracy in data extraction modules, with F-measures > 0.98.
• Synthetic trial results showed similar directionality and magnitude to the comparator clinical trial.

Conclusions:

• The developed framework provides an accurate and user-friendly method for EMR data extraction.
• This approach enables the creation of synthetic clinical trials to complement traditional research.
• Further validation in diverse patient cohorts is needed to optimize patient care and drug development.