A method for the graphical modeling of relative temporal constraints

Area of Science:

• Biomedical Informatics
• Health Data Science
• Clinical Research Informatics

Background:

• Electronic patient data analysis requires defining temporal relationships between criteria.
• Current form-based methods struggle with complex temporal constraints in translational research.
• A graphical, accessible method for temporal query modeling is needed.

Purpose of the Study:

• To develop a new graphical notation for creating complex temporal queries.
• To enable easier and more implementable temporal query modeling.
• To improve patient cohort identification in electronic health records.

Main Methods:

• Developed a graphical notation based on Allen's time interval algebra.
• Extended the notation with duration intervals and time interval modifiers.
• Created a prototypical implementation with a frontend and i2b2-connected backend.
• Generated database queries from the graphical notation.

Main Results:

• The graphical notation effectively models temporal queries using symbolic time intervals.
• Extensions allow inference of temporal distances and event counting/exclusion.
• Prototypical implementation demonstrated feasibility on the MIMIC-III database.
• Successfully modeled typical temporal phenotyping queries.

Conclusions:

• The proposed graphical notation offers an accessible and implementable solution for complex temporal query modeling.
• This approach enhances patient cohort identification in translational research.
• The method is suitable for real-world applications using databases like MIMIC-III.