Sponsors
The bill aims to improve medication adherence and lower drug spending through AI-driven outreach and greater use of generics/biosimilars, but it risks privacy harms, increased administrative costs, and care complications if strict adherence targets or substitution pressures are enforced without strong safeguards.
Medicare beneficiaries with chronic conditions are likely to have better medication-taking (higher adherence), which can reduce treatment failures and hospitalizations.
Medicare beneficiaries and taxpayers may pay less for drugs as the bill encourages greater use of lower-cost generics and biosimilars.
Hospitals, health systems, and state programs can run more efficient adherence outreach and monitoring by using AI/ML to target interventions, potentially saving staff time and resources.
Patients with chronic conditions and their providers may face burdensome monitoring or quasi-prior-authorization controls if a 90% adherence target is enforced, harming care flexibility.
Medicare beneficiaries face increased privacy risks because AI-driven adherence tools will process sensitive medication and health data that could be shared or exposed.
Some patients may lose timely access to needed brand-name therapies if guidance pressures substitution to generics/biosimilars without adequate clinical exceptions.
Based on analysis of 2 sections of legislative text.
Requires HHS to issue AI/ML-based drug adherence guidelines aiming for 90% adherence for Medicare Part B and Part D drugs and to promote generics/biosimilars when practicable.
Requires the HHS Secretary to create national drug adherence guidelines for Medicare Part B and Part D drugs that aim for 90% patient adherence. The guidelines must use artificial intelligence and machine learning tools and, when practicable, encourage use of generic and biosimilar drugs. The bill defines which drugs are covered by reference to existing Medicare Part B and Part D definitions but does not specify funding, enforcement mechanisms, or a timetable for implementation.
Introduced February 7, 2025 by David Schweikert · Last progress February 7, 2025