Ensuring Safe AI Use in Healthcare
Adopting artificial intelligence (AI) in healthcare comes with significant responsibilities, particularly in establishing a solid governance framework. The Health Sector Coordinating Council (HSCC) recently released guidance emphasizing the risks of algorithm failures, biases, and the complexities associated with protecting patient information as mandated by HIPAA.
This comprehensive 87-page document is part of an ongoing series focused on AI-related cybersecurity, developed by the HSCC’s special task group on AI governance. This group plans to continue publishing guidance, having already addressed risks associated with third-party AI providers. The authors of the report hail from prominent health systems and tech companies throughout the nation.
The guidance highlights that effective AI governance should weave cybersecurity principles into every stage of AI system management, including design, deployment, and eventual retirement. It outlines essential protocols for safe data management, protecting AI models, detecting threats, and continuously monitoring for risks such as data leaks and model vulnerabilities.
However, the HSCC cautioned that this guidance should not be used in isolation. Organizations need to integrate it with their existing governance practices for a comprehensive approach.
Topics covered in the document include clinical safety, ethical considerations, cybersecurity measures, risks from generative AI and large language models, supply chain concerns, and planning for AI-related incidents.
The report warns that without proper governance, AI systems may risk data breaches, operational disruptions, and may even compromise patient safety, resulting in harm and a tarnished reputation for healthcare organizations.
To address these challenges, the HSCC advises that organizations form an AI cyber governance committee. This committee should include diverse members such as program leaders, healthcare professionals, IT and security teams, legal advisors, and representatives for patients.
Overall, the guidance underscores the complexities of safe AI implementation in healthcare. Organizations need to account for not only cybersecurity risks but also the implications for operations and patient care.
The HSCC emphasizes that, given the evolving nature of healthcare, managing AI effectively is essential for ensuring patient safety.
Jill Hughes has been reporting on health technology since 2021, focusing on cybersecurity, HIPAA compliance, interoperability, AI, and electronic health records (EHRs).
