Epic to Launch AI Validation Software to Enable Health Organization Assess AI Model’s Accuracy

The new AI trust and assurance software suite automates data collection and mapping, providing near real-time metrics and analysis on AI models.

Electronic health records giant Epic has developed an AI validation software suite aimed at empowering healthcare organizations to assess their AI models' accuracy and performance locally as well as monitor their performance over time, Fierce Healthcare reported.

Epic’s senior VP of Research and Development, Seth Hain, revealed that the launch is in response to the need for robust validation standards as AI continues to transform healthcare.

The new AI trust and assurance software suite automates data collection and mapping, providing near real-time metrics and analysis on AI models.

This automation streamlines the validation process, eliminating the need for healthcare organizations' data scientists to manually map data, which is often the most time-consuming aspect of validation, Hain elaborated.

He also emphasized the importance of enabling AI testing and validation at a local level while allowing for ongoing monitoring at scale.

What does Epic’s AI Validation Software Offer?

Hain stated, "We'll provide health systems with the ability to combine their local information about the outcomes around their workflows, alongside the information about the AI models that they're using.”

“They will be able to use that both for evaluation and then importantly, ongoing monitoring of those models in their local contexts," he added.

The software suite includes intuitive reporting dashboards updated automatically, offering analysis broken down by demographics such as age, sex, and race/ethnicity.

Additionally, it features a common monitoring template and data schema to facilitate the extension of the suite to new AI models in the future.

One of the key aspects of Epic's initiative is its commitment to making the suite's monitoring template and data schema publicly available, enabling healthcare organizations to monitor their own custom AI models as well as those from third-party vendors.

"At scale, [the suite] opens the opportunity for health systems to be able to understand outcomes alongside AI, and do so in a way that is open and flexible to the evolving best practices around that type of analysis."

Hain underscored that as AI best practices evolve, the open-source framework will allow organizations to incorporate these standards and practices into their analyses.

Need for AI Validation Software for Healthcare

Epic plans to release the capability within the next four to six weeks, with ongoing updates throughout the summer.

Hain believes that tools like the AI validation software suite will play a crucial role in building trust in healthcare AI by providing a framework for analyzing and understanding outcomes alongside AI technology.

For nearly ten years, healthcare organizations have employed predictive AI models and machine learning. However, the emergence of large language models (LLMs) and generative AI tools introduces a distinct challenge.

Health systems are rapidly adopting LLMs and generative AI to address tasks such as medical record summarization and clinical note automation.

Nonetheless, these pioneers are still refining optimal methods to validate AI models, aiming to enhance confidence in their accuracy, performance, and safety.

Epic’s move to launch AI Validation software could prove to be a step in the right direction to provide tools that enable local auditing of AI models.

Ferrum Health’s AI Validation Software

Last year, Ferrum Health became the first to launch a comprehensive healthcare AI platform on the Oracle cloud marketplace.

It offers AI validation by leveraging model operations in conjunction with GPT to automatically validate AI applications using hospitals' patient data.

The system analyzes radiology reports using GPT to establish ground truth diagnoses and flags discrepancies between language and visual classifiers for review, providing hospitals with insights into AI tool performance.

Its AI validation software addresses the limitations of current validation methods, streamlining the process for hospitals and ensuring that AI algorithms provide a clinical benefit.