Discovering associations between adverse drug events using pattern structures and ontologies.

Area of Science:

• Computational biology
• Health informatics
• Medical data mining

Background:

• Patient data, including electronic health records and adverse event reporting systems, are crucial for studying Adverse Drug Events (ADEs).
• Identifying frequently associated ADEs within specific patient subgroups presents a significant challenge in pharmacovigilance.

Purpose of the Study:

• To explore an original approach for identifying frequently associated ADEs in patient subgroups.
• To develop a flexible and expressive representation of patient ADEs.

Main Methods:

• Utilizing formal concept analysis and its pattern structures, a mathematical framework for generalization.
• Integrating domain knowledge from medical ontologies to enhance analysis.
• Applying the approach to two distinct datasets across three different settings.

Main Results:

• Demonstrated the flexibility of the approach in extracting association rules at various levels of generalization.
• Successfully identified distinct ADEs that co-occur within specific patient groups.
• Validated the approach's efficacy across multiple datasets and analytical settings.

Conclusions:

• The proposed method offers an expressive way to represent patient ADEs.
• Extracted association rules can form the basis for a recommendation system for ADE management.
• The representation is adaptable and can incorporate additional ontologies and diverse patient records for expanded insights.