Researchers at the AI for Health Institute have created a new framework designed to enhance the collaboration between artificial intelligence and human clinicians—an essential challenge in the realm of healthcare AI today.
By merging clinical insights with the extensive and rapidly expanding knowledge base of AI, there is immense potential for revolutionizing healthcare practices. This synergy can lead to earlier diagnoses and improved outcome predictions. However, current AI systems often face dangers stemming from inaccuracies or unwarranted confidence in their predictions.
To tackle this issue, Sizhe Wang, a graduate student working under Chenyang Lu, the Fullgraf Professor at WashU McKelvey Engineering, has developed a model known as Clinical Uncertainty Risk Alignment (CURA). This innovative framework instructs clinical AI on when to exhibit confidence and when to exercise caution, equipping it to deliver more reliable estimates regarding certainty and uncertainty surrounding its predictions. The CURA model is scheduled for presentation at the upcoming Association for Computational Linguistics annual meeting in July.
For additional insights, visit the McKelvey Engineering website.
