Oct. 21, 2024 — Gesund.ai, the clinical AI validation company, has announced the open sourcing of a comprehensive library of evaluation methodologies and frameworks to help developers and users address AI performance and potential bias across patient cohorts for the FDA-regulated AI solutions on the market. As part of its mission to standardize AI validation, Gesund.ai also announced it has joined the Coalition for Health AI, after partnering with public and private industry groups like CancerX, the NIST AI Safety Institute Consortium, VALID AI and the American Heart Association.
Clinical AI products have special characteristics that must be accounted for when screening them for potential bias. For instance, beyond the data itself, the laws themselves are adapting and evolving. In 2025, healthcare providers and payors and not just AI developers will be responsible for ensuring discrimination does not occur from the use of clinical algorithms and other AI solutions.
While the FDA has its own guidelines for clinical AI performance testing, there is no industry standard platform for doctors or developers to use on their own. Gesund.ai has spent years creating a clinical AI validation software suite now used across different healthcare settings. It encompasses many different FDA-aligned feature sets relevant for real-world deployment. Now, the company is opening up the library of methodologies and frameworks used in combination with its validation engine for anyone to use, update and edit.
“One goal of open sourcing this library is to rally communities together to update, improve and expand standards and methodologies as the field evolves,” said Dr. Enes Hosgor, CEO and founder of Gesund.ai. “No one company can or should do it alone.”
Called Gesund Validate, Healthcare, AI and industry professionals can access it at this link. Validate is open source and has many benefits for AI developers and healthcare providers, like:
- Covers the clinical AI solutions physicians use, like radiology, because it is built on unstructured multimodal data adopting “expert-in-the-loop” workflows and a wide range of computer vision tasks.
- Spots bias by analyzing model performance across patient metadata like age, gender, and race against built-in fairness and bias mitigation methodologies.
- FDA guidelines are the core of Gesund Validate and it can be adapted per the unique needs of a given hospital as well as to accommodate evolving laws.
To learn more visit www.gesund.ai.
June 04, 2025 
