An approach to knowledge base construction based on expert opinions.

Area of Science:

• Neurology
• Epileptology
• Medical Informatics

Background:

• Knowledge bases are crucial for medical research and clinical decision-making.
• Expert opinions are a valuable source for constructing specialized medical knowledge bases.
• Validating expert-derived data is essential for ensuring knowledge base accuracy.

Purpose of the Study:

• To present and validate a method for constructing a knowledge base using expert opinions.
• To assess the reliability of expert-derived data on epileptic seizure manifestations.
• To demonstrate an approach for identifying consistent and dissenting expert viewpoints.

Main Methods:

• Constructed a knowledge base on epileptic seizure manifestation frequencies using data from neurologists/epileptologists.
• Utilized the inter-rater intraclass correlation coefficient (ICC) to measure response reliability.
• Employed the Spearman-Brown prophecy formula to determine the necessary number of additional experts.
• Developed methods to assess expert data consistency and identify deviating opinions.

Main Results:

• Achieved an ICC close to 0.9 for all seizure types after expert inclusion, meeting the target reliability.
• Successfully identified one expert with diverging opinions, whose exclusion improved overall data reliability.
• Demonstrated good statistical power (>=0.75) for the developed assessment methods.
• Validated the reliability of expert-provided data on epileptic seizure manifestations.

Conclusions:

• Human experts can reliably provide data on the frequency of epileptic seizure manifestations.
• The developed approach effectively distinguishes between consistent and diverging expert opinions.
• This methodology facilitates the construction of robust and reliable medical knowledge bases.