Maverick’s Monthly: July 2025 Snapshot

In this newsletter, we review two themes from the news that are worth a second look. This month’s edition highlights two areas of focus in the Trump Administration’s HHS:

1. Fraud, waste, and abuse (FWA) prevention in federal health programs, and
2. Implementing an Artificial Intelligence (AI) strategy.

How has the government led “fraud, waste, and abuse” efforts in federal health programs?

Maverick’s flags about the FWA efforts to date:

• CMS
  • CMS finalized its ACA program integrity rule, which streamlines enrollment periods and tightens eligibility verification to address fraud and improper enrollments. The agency projects this policy will reduce marketplace premiums by 5% and generate $12B in savings, though it estimates up to 1.8M individuals could lose coverage.
  • The CMS Innovation Center announced a new model for traditional Medicare that applies tech-enabled prior authorization processes to target fraud and abuse. (See here for Maverick’s analysis of the WISeR Model)
  • The CMS Fraud Detection Operation Center (FDOC) continued expanding its analytical capabilities.
• DOJ
  • The DOJ and HHS announced two new healthcare fraud initiatives: a Data Fusion Center and a False Claims Act (FCA) working group. The announcement coincided with the results of the DOJ’s largest healthcare fraud takedown in history, charging 324 people in schemes totaling over $14.6B.
    • The Data Fusion Center will use AI and cloud-computing to enhance detection and prosecution efficiency for healthcare fraud schemes.
    • The FCA working group will focus on Medicare Advantage oversight, drug/device/biologics pricing models, barriers to access and services covered by federal programs, defective equipment, and electronic health record manipulation.

• Industry
  • Doctors for America, the Main Street Alliance, and three city governments filed a lawsuit against CMS, alleging that the ACA program integrity rule creates substantial enrollment barriers that will lead to unnecessary coverage loss rather than meaningful fraud prevention.
  • HL7, a nonprofit healthcare standards organization, unveiled a new AI office with initiatives related to AI explainability, safe clinical deployment, and healthcare fraud detection.
    • In May, HL7 released a landscape assessment of how AI can improve fraud detection and payment integrity. The report included three immediate recommendations: responsible AI standards development, pilot programs to foster payer-provider collaboration, and alignment with existing federal and state regulatory frameworks.

The federal government is making AI a top priority, but what does this look like in practice across HHS?

How HHS subagencies are implementing AI:

• FDA
  • AI Leadership: Jeremy Walsh, Chief AI Officer; Vinay Prasad, Chief Medical Officer and Center for Biologics and Evaluation (CBER) Director; Anindita (Annie) Saha, Center for Drug Evaluation and Research (CDER) AI Policy Lead and Digital Health Center of Excellence Associate Director.
  • Recent AI Actions:
    • Elsa – An LLM built on Anthropic’s Claude that has been integrated into FDA workflows to help accelerate regulatory review timelines. Walsh estimated nearly 6,000 staff use Elsa weekly for tasks like application review and adverse event report redaction, with expanded capabilities planned for the future.
    • “Priorities for a new FDA” – Commissioner Makary and Vinay Prasad described the agency’s first AI-assisted scientific review pilot, plans to modernize the review process for AI-based technologies, and a road map to reduce animal testing using AI-based computational modeling in a JAMA Viewpoint
    • Cross-Agency AI Councils – The FDA will reportedly launch two AI councils to consolidate AI efforts with one focused on internal implementation and another on policy development.
• CMS
• • AI Leadership: Andrea Fletcher, Chief Digital Strategy Officer; Jacob Shiff, CMS Innovation Center Chief AI and Technology Officer.
  • Recent AI Actions:
    • Physician Fee Schedule & Outpatient Prospective Payment Schedule proposed rules – CMS asked for input from the public on how better to account for the costs and use of SaaS and AI technologies. CMS also included AI-specific proposals in the payment rules, such as expanding patient safety reporting to capture cases where AI may have contributed to actual or potential harm and seeking feedback on how AI improvements to service delivery might require payment model adjustments.

• NIH
  • AI Leadership: The NIH’s recent AI initiatives were led by Gil Alterovitz, who briefly served as interim NIH Chief AI Office before returning to his current role as Chief AI Officer at the Department of Veterans Affairs
  • Recent AI Actions:
    • AI Strategic Plan – On June 3, the NIH released a request for feedback (RFI) seeking feedback on an institute-wide AI strategy that will help transition from current data-driven tools to advanced, autonomous AI systems in biomedical research and healthcare. The RFI asks for input on strategic priorities, governance models, and actionable steps for the first year of implementation.
    • Ongoing Efforts – The NIH Office of Data Science Strategy is leading three institute-wide initiatives related to responsible AI in biomedical research, ensuring ethical and transparent research, and promoting alignment with FAIR (Findable, Accessible, Interoperable, Reusable) principles.

• CDC
  • AI Leadership: Travis Hoppe, Chief AI Officer
  • Recent AI Actions:
    • According to its data modernization webpage, the CDC is currently using AI/ML for: disease diagnosis improvement, data quality enhancement, health trend identification, and equity and transparency promotion.
    • The CDC is exploring future applications of AI/ML for: opioid and overdose crisis forecasting, foodborne outbreak tracking, safe vaccine delivery assistance, and sensitive patient data identification from unstructured text sources.
      • Hoppe said he wants to expand CDC industry partnerships with AI innovators to explore deep research and agentic AI in a LinkedIn post announcing his new role as Chief AI Officer.
    • MedCoder – The National Vital Statistics System completed full deployment of a ML and natural language processing system that automates death certificate cause coding. The platform processes nearly 90% of mortality records without human intervention, compared to the previous system’s 75% capacity.

Maverick’s Monthly will spotlight two more themes next month; please send a note to julie.barnes@maverickhealthpolicy.com if you have requests about focus areas.