Epic has launched its first open-source tool, an AI validation software suite, to help healthcare organizations test and monitor artificial intelligence models. This tool, available for free on GitHub, allows health systems to validate AI models integrated with electronic health records (EHRs), including those developed by Epic and other organizations. By automating data collection and mapping, it provides near real-time metrics and analysis, aiming to ensure AI equity and enable local-level AI testing and ongoing monitoring.
The tool is designed to enhance the use of AI in healthcare by offering intuitive reporting dashboards that break down data by demographics such as age, sex, and race/ethnicity. This helps ensure AI models work fairly across different patient groups. The software suite includes a common monitoring template and data schema to facilitate extending its capabilities to new AI models in the future. Epic plans to expand the tool's validation capabilities to generative AI models.
The Health AI Partnership (HAIP), including organizations like Duke Health and Mayo Clinic, plans to use Epic's tool for local validation of AI models. Epic also plans to collaborate with HAIP and the University of Wisconsin to study predictive models. Despite concerns about potential conflicts of interest, Epic asserts the tool is designed to work with any predictive models, emphasizing its commitment to AI equity and the responsible use of AI in healthcare.
Click here to read the original news story.