HHS releases AI strategy, united by new OneHHS approach

The Department of Health and Human Services has released a new strategy to apply artificial intelligence agencywide, implementing the technology to enable data sharing, scientific research and expedited agency operations.

Announced on Thursday, the HHS AI Strategy highlights five pillars where AI will be incorporated: ensuring governance and risk management; designing infrastructure for user needs; promoting workforce development and burden reduction; fostering research through gold-standard science; and enabling public health delivery modernization. 

“For too long, our Department has been bogged down by bureaucracy and busy-work; even the most productive public servants are mired in paperwork and process. Across our mission space — from research, to health care delivery, to public health — we’ve layered administrative red tape on innovators,” HHS Deputy Secretary Jim O'Neill wrote in an introductory letter for the strategy. “We will harness AI technologies to streamline operations and enhance support for care delivery throughout the entire health care industry. We call it OneHHS.”

Within the first pillar, HHS is prioritizing instilling public trust in its AI usage. This will entail developing a new governance structure to oversee AI projects’ implementation and incorporating “risk-proportionate controls” to ensure system integrity. AI impact assessments, independent reviews and a well-maintained use case inventory are all included in this effort.

The strategy also introduces the “OneHHS” approach as the starring component of the second pillar, and it will serve to connect subagencies and offices to better share data with one another. HHS noted that data and knowledge sharing with the public will occur where it is “legally permissible.”

“By building this OneHHS AI-integrated Commons, the Department will ensure that new AI solutions can be developed, tested, and deployed rapidly, with the ability to operate in different environments and across different systems,” the strategy reads. 

To maintain adequate security amid enhanced data sharing, HHS is finalizing policies and directives to overhaul ownership of HHS data internally versus from industry partners, standardizing data sharing governance that will inform AI model use and performance related to internal decisionmaking, and curating data to ensure it is in a format that is legible for new AI and machine learning tools. 

The third pillar aims to promote workforce development while expediting normal HHS business operations. As HHS begins to implement AI solutions, the strategy says the agency “envisions a work environment where AI is seamlessly integrated into operations, augmenting human capabilities, increasing productivity, and improving job satisfaction, all while maintaining the high standards of quality, accuracy, and accountability that HHS’s mission demands.”

This will include establishing training pathways aligned to employee roles. It will also involve deploying secure and approved AI tools into existing HHS systems and data, creating repositories for AI prompts and common tasks, and retaining and attracting top AI and data talent to support these endeavors. 

Pillar four aims to ensure AI tools have established benchmarks to gauge their efficacy in accelerating scientific discovery. HHS will develop new, centralized approaches for research and development when using AI, promoting open-weight and open-source models to be used in safe capacities, as well as creating reproducible pipelines to expedite the validation of AI for research efforts.

The final pillar focuses on ensuring that the public benefits from AI implementation within HHS, particularly in a diagnostic capacity. The strategy underscores that, as with internal employees, AI is not intended to replace physicians and health care workers, but to support their existing roles. 

Aside from using AI in diagnostics and preventative analytics, pillar five aims to help centrally coordinate HHS office roles in advancing AI in health care and support the continuous monitoring of implemented AI software. 

Following the implementation of the strategy, HHS will both develop specific metrics that measure if each pillar is meeting its strategic goals and publish the results.