Cohort selection for clinical trials: n2c2 2018 shared task track 1

Area of Science:

• Medical Informatics
• Natural Language Processing
• Clinical Research

Background:

• The 2018 National NLP Clinical Challenges focused on patient cohort identification from medical records.
• Accurate patient selection is crucial for clinical trial recruitment and research.

Purpose of the Study:

• To evaluate natural language processing (NLP) system performance in identifying patients meeting specific clinical trial criteria.
• To analyze various NLP approaches for extracting patient data from longitudinal medical records.

Main Methods:

• Annotated 288 American English clinical narratives for patient eligibility based on trial criteria.
• Included criteria requiring concept extraction, temporal reasoning, and inference.
• 47 teams participated, submitting 109 system outputs.

Main Results:

• The top-performing system achieved a micro F1 score of 0.91.
• Rule-based and hybrid systems dominated the top 10 rankings.
• Consulting medical professionals improved system recall.

Conclusions:

• No single NLP solution fits all patient identification tasks.
• Future NLP research should address complex inferences, temporal reasoning, and domain knowledge integration.
• Interpreting clinical narratives requires careful consideration of annotator domain knowledge.