Dedicated to systematically improving the safety, reliability, and equity of artificial intelligence in healthcare, primarily focused on medical imaging.
The clinical impact of AI begins only after it crosses the ‘trust threshold.’ The AIDE Lab helps clinical AI earn that trust.
David Larson, MD, MBA Co-Director, AIDE Lab Professor, Radiology
The AIDE Lab develops the tools and frameworks necessary to assess how AI algorithms add value to physicians, hospitals, and most importantly, patients.
Akshay Chaudhari, PhD Co-Director, AIDE Lab Assistant Professor, Radiology
The clinical impact of AI begins only after it crosses the ‘trust threshold.’ The AIDE Lab helps clinical AI earn that trust.
David Larson, MD, MBA Co-Director, AIDE Lab Professor, Radiology
The AIDE lab develops the tools and frameworks necessary to assess how AI algorithms add value to physicians, the hospital, and most importantly, the patients.
Akshay Chaudhari, PhD Co-Director, AIDE Lab Assistant Professor, Radiology
Research Areas
Develop AI applications that improve the quality and effectiveness of imaging-based healthcare
Algorithm Development
leverage machine learning techniques to address inefficiencies and challenges in maintaining high quality in imaging-based healthcare
Evaluation Methods
comprehensively assess AI by focusing on real-world clinical variables to understand model capabilities and limitations
Quality Assessment
develop strategies and recommendations to conduct robust quality assessments of AI models
AI Monitoring
build tools to detect anomalies, performance deviations, and potential risks associated with AI deployment
Clinical Translation
Help ensure safe and reliable performance of radiology-related AI applications
Pre-deployment
evaluate and curate AI tools to ensure that radiologists are getting high-quality applications that directly impact their work
Implementation
facilitate large-scale testing with local data to ensure performance and perform pre-work for subsequent monitoring
Post-deployment
measure expected benefits, detect and analyze new failure modes, and recommend actions to correct for performance deviations