Objective:

To evaluate the effectiveness of an AI system in identifying eligible patients for clinical trials in a rare heart condition.

Key Findings:

• 37% of AI-identified patients were Black, compared to 7% identified manually.
• 60% of AI-identified patients were not linked to a cardiologist, indicating broader outreach.
• The AI system correctly excluded ineligible patients 99% of the time.

Interpretation:

The AI system may enhance diversity in clinical trial recruitment and streamline the patient identification process, potentially reducing bias.

Limitations:

• Further evaluation is needed to compare AI-assisted screening with conventional methods.

Conclusion:

AI systems could significantly reduce the workload of manual chart reviews while improving patient identification efficiency for complex clinical studies.