AI BriefingBuilding a Pioneering Model for AI Clinical Copilots with Penda Health
Key point
A study by OpenAI and Penda Health found that adopting an AI clinical copilot reduced diagnostic errors by 16% and treatment errors by 13%.
Details
While LLM performance is advancing rapidly, its adoption in real-world healthcare settings remains slow. To close the model-implementation gap—the gap between a model's capabilities and its actual use—OpenAI partnered with Penda Health, a primary care provider in Kenya.
AI Consult, developed by Penda Health, serves as a real-time safety net in clinical settings. This copilot provides LLM-based recommendations at key moments during patient visits, and is designed to help prevent errors while ensuring clinicians remain in control.
In a study covering approximately 40,000 patient visits, clinicians who used AI Consult achieved a relative reduction of 16% in diagnostic errors and 13% in treatment errors compared to those who did not.
These results were driven by three key factors:
- Strong model performance: Built on a high-performance model powered by GPT-4o.
- Clinically optimized implementation: Co-developed with clinicians to avoid disrupting the flow of care.
- Proactive field deployment: Training and support to help clinicians understand and effectively use the copilot.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.