Lecture Notes: Dr. Lindsay Knaik on Epic Cosmos

Introduction of Dr. Lindsay Knaik

• Dr. Lindsay Knaik from the University of Iowa
• Professor of Metrics and Associate CMIO at Stead Family Children's Hospital
• Educational background in Biomedical Engineering and Medical School in Iowa
• Residency at Baylor, Fellowship in Neonatology at Vanderbilt
• Known for advanced studies in neonatal care and informatics

Overview of Cosmos

• Research-related sharing network
• Provides a large, de-identified database from Epic institutions
• Contains over 203 million patients and billions of encounters
• Designed to facilitate research with clinical EHR information

Advantages of Cosmos

• One of the largest databases for clinical EHR data
• Fast query capabilities
• Representative of U.S. demographics, similar to the U.S. Census
• Includes diverse datasets: labs, meds, social determinants of health

Challenges and Considerations

• Longitudinal data may have gaps due to non-Epic organizations
• Data reliability and completeness are crucial
• Diverse data set with some limitations (e.g., hourly flow sheet data)

Accessing Cosmos

• Requires belonging to an Epic institution
• Legal terms and internal advocacy play a role in access
• Training required for users (Epic classes and data science tools)

Using Cosmos: Examples and Tools

• Examples of neonatal hypertension research
• Slicer Dicer for initial cohort creation
• Data Science side for more detailed, line-level data
• Epic dashboards and data science environment tools

Challenges in Data Usage

• Translating Slicer Dicer queries to data science environment
• Limitations in medication data accuracy
• Sidekick: a beta tool for large language model assistance

Working with the Epic Cosmos Team

• Open to feedback and improvements
• Epic research blog for updates on findings
• Collaborations with universities for validation and study

Future of Cosmos

• Potential for large consortium databases
• Continuous improvements in data quality and capabilities
• The impact of trained multidisciplinary teams for research

Conclusion

• Cosmos offers significant potential for large-scale research
• Continued exploration and validation of data is essential
• Useful tool for understanding trends and outcomes in healthcare