Computational Framework for Structuring and Analyzing Clinical Trial Criteria for AI-Guided Fine-grained Matching

Area of Science:

• Clinical Informatics
• Artificial Intelligence in Healthcare
• Clinical Trial Management

Background:

• Current artificial intelligence (AI) solutions for clinical trial matching often use oversimplified criteria or limited structured data.
• This limitation hinders the potential of AI to effectively automate and optimize patient recruitment for clinical trials.
• Existing methods struggle with the inherent complexity and variability of clinical trial eligibility criteria.

Purpose of the Study:

• To introduce a novel framework for structuring and analyzing eligibility criteria from real-world clinical trial protocols.
• To inform the development of more granular and effective AI-driven clinical trial matching strategies.
• To quantify the complexity and variability of clinical trial eligibility criteria.

Main Methods:

• Eligibility criteria from three clinical trial protocols were systematically decomposed into individual variables.
• Variables were evaluated based on data type, scope, and interdependencies.
• A novel complexity formula, incorporating variable interdependence and Flesch-Kincaid reading grade level, was developed and applied.

Main Results:

• Clinical trial protocols contained between 22 and 160 eligibility variables, with 4-22% exhibiting interdependence.
• Reading grade levels for criteria ranged from sixth grade to first-year college, indicating significant accessibility variation.
• Complexity scores varied substantially, with high cognitive and logical burdens observed in some protocols, often featuring recursive and hierarchical structures.

Conclusions:

• Clinical trial eligibility criteria exhibit significant variability and structural complexity, posing challenges for current AI matching systems.
• A standardized method for assessing trial complexity can enhance AI algorithm transparency, scalability, and interpretability.
• Developing structured, computable frameworks is crucial for improving the equity and efficiency of clinical trial recruitment.