A comparison of the temporal expressiveness of three database query methods

Area of Science:

• Health Informatics
• Clinical Decision Support Systems
• Database Design

Background:

• Time modeling in clinical decision-support systems presents design challenges.
• The necessity of interval semantics in clinical database models remains unresolved.
• Querying protocol-based conditions requires effective temporal data manipulation.

Purpose of the Study:

• To investigate the requirement for interval-based operations in querying protocol-based conditions.
• To compare the expressiveness of Arden Syntax, SQL, and TimeLineSQL (TLSQL) for temporal querying.
• To evaluate the suitability of different temporal query methods for clinical trial eligibility criteria.

Main Methods:

• Analysis of 256 eligibility criteria from the T-HELPER system for HIV clinical trials.
• Comparison of three temporal querying methods: Arden Syntax, SQL, and TLSQL.
• Evaluation of the ability of each method to express temporal conditions, including interval-based patterns.

Main Results:

• 79% of criteria involve timestamp operations, with four distinct temporal patterns identified.
• Arden Syntax handles two non-interval patterns (54% of temporal conditions).
• SQL handles two non-interval and one interval pattern (96% of temporal conditions).
• TLSQL, supporting interval semantics, expresses all four temporal patterns.

Conclusions:

• The T-HELPER system necessitates simple temporal operations for most protocol-based queries.
• TLSQL demonstrates superior expressiveness for the temporal conditions encountered in this system.
• Interval-based temporal modeling is crucial for comprehensive querying of clinical trial eligibility criteria.