Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Technology Implementation topic
No spam. Unsubscribe anytime.
Committee chair presses experts on moving AI and data tools from ideas to agency implementation
Summary
At a Joint Economic Committee hearing, panelists urged tangible steps to move AI and analytics into federal agencies, recommending a centralized analytics hub, automation of Medicaid eligibility, uniform diagnosis coding across Medicare, and electronic prior-authorization systems.
A Joint Economic Committee hearing on technology and health policy focused on how to turn reports and pilots into agency action, with lawmakers and experts identifying workforce gaps, legal constraints and concrete tech fixes.
The committee chair opened by asking how to “take your technology, your understanding of the technology” and move it into agencies so they stop producing ideas and start implementing them. He noted legacy systems such as the AS/400 and said modern ‘‘translational’’ tools can convert older code to current languages, but agencies repeatedly fall short on execution.
Dr. Thomas, speaking as an expert witness, said two central problems recur: workforce capacity and the difficulty of integrating tools across siloed agency databases. He proposed a centralized analytics or "AI shop" that would host agency datasets with role-based permissions and provide shared machine-learning tools. "The technology is there," he said, but Congress-defined privacy rules and agency incentives often block practical rollout.
Dr. Miller urged three near-term priorities for legislation or executive action to lower costs and improve access to care using technology. First, he recommended automating Medicaid eligibility — moving from paper and fax-based processes to an electronic, data-matching system that pulls state records and, where appropriate, vetted commercial data to speed determinations. Second, he advocated standardizing diagnosis coding across Medicare fee-for-service and Medicare Advantage so beneficiaries are coded consistently and incentives to ‘‘score’’ patients sicker are reduced. Third, he called for making prior-authorization data submission electronic and integrated at the point of care, not a quarterly or piecemeal reporting exercise.
On risk adjustment and procurement, panelists discussed getting coding profiles in front of payers before capitated bids so plans bid on care rather than on coding discrepancies. "If you fix coding across the entire system ... it would eliminate those incentives," Dr. Miller said.
Witnesses also emphasized fraud analytics as a ‘‘living’’ dataset that must be continuously updated; they compared the model to cybersecurity threat-sharing, where new signatures discovered by one agency are shared broadly so others can adapt. One proposal offered as a starting point was a centralized analytics center, similar in spirit to the Program of All-Inclusive Care for the Elderly (PACE) model in its centralized approach to coordination.
The committee did not take votes or adopt formal motions during the session. The chair closed by thanking the witnesses, suggesting a whiteboard follow-up to sketch implementation steps, and adjourning the hearing.
The hearing highlighted a consistent theme: tools exist to modernize government IT and health-data workflows, but implementation will require legislative or executive direction, funding for agency talent, and careful design around privacy and data-sharing rules.
