Enabling Safe AI

In January, CHS announced a partnership with Google Cloud to unify data from its affiliates into a single platform that facilitates greater transparency, enables more real-time decision-making and that will serve as the foundation for the future use of AI in our health care settings. By leveraging AI, we aim to drive efficiencies that allow our health care professionals to focus more of their time on high-value patient interactions. The potential of AI in healthcare is boundless, and we are just beginning to explore its transformative capabilities.

To ensure the responsible implementation of AI, we are actively implementing several measures:

• Data Diversity: We prioritize collecting diverse and representative datasets to ensure that AI models are trained on a wide range of patient characteristics and conditions, mitigating the risk of bias.
• Algorithmic Auditing: We regularly conduct algorithmic audits to identify and address potential biases in our AI models. This process involves analyzing model predictions across different subgroups to identify any disparities or unfair outcomes.
• Human Oversight: We maintain robust human oversight mechanisms to ensure that AI-driven decisions are always subject to human review and intervention when necessary. This safeguard prevents AI systems from making critical decisions without human input.
• Continuous Learning: We stay updated with the latest research and best practices in AI bias mitigation. Our commitment to continuous learning allows us to continually refine our approach and adopt cutting-edge techniques to minimize AI bias.
• Collaboration and Partnerships: We actively engage in collaborations and partnerships to gain diverse perspectives and insights into AI bias reduction strategies.

One such collaboration is with the Coalition for Health AI (CHAI), an influential consortium dedicated to shaping safe and responsible AI practices in healthcare. By collaborating with CHAI, we collectively work to align standards and reporting for health AI, educate end-users on technology evaluation, and drive adoption in a manner that prioritizes health equity and addresses algorithmic bias.

CHAI achieved a noteworthy milestone last April with the release of its highly anticipated “Blueprint for Trustworthy AI Implementation Guidance and Assurance for Healthcare.”

This comprehensive 24-page guide provides timely recommendations to enhance trustworthiness, ensure high-quality care, and effectively address healthcare needs in the rapidly evolving landscape of health AI tools. The blueprint is a product of the collaborative efforts of leading experts from academic medical centers, healthcare, technology, and other industries, working closely under the observation of federal agencies throughout the past year.

“Joining a consortium like CHAI is imperative,” said Dr. Ashwin Prakash, CHS Vice President of Clinical Data Science. “By collaborating with industry peers, sharing best practices, and influencing regulatory policies, coalition members play a crucial role in advancing the responsible and ethical use of AI in healthcare. This participation offers tangible business benefits, including enhanced market competitiveness, risk mitigation, and organizational learning,” he concluded.