FDA Policy/General

Gottlieb Urges FDA Regulatory Pathway for ‘Clinical AI’

Former FDA commissioner Scott Gottlieb is calling on Congress to establish a new regulatory pathway for clinical artificial intelligence models developed and used within hospitals, warning that the absence of consistent federal oversight could undermine patient safety and limit broader commercialization of clinical AI technologies.

In an October 8 commentary published in JAMA Health Forum, Gottlieb argued that clinical AI is following a path similar to laboratory-developed tests (LDTs), which historically operated mainly outside FDA premarket review.

“Clinical AI tools are proliferating in academic centers,” Gottlieb wrote. “Duke Health built a deep learning system for predicting sepsis and moved it into routine care. Johns Hopkins University tested its Targeted Real-Time Early Warning System for detecting sepsis across 5 hospitals. At NYU Langone Health, a language model trained on the institution’s clinical notes predicts readmissions. In each case, the institution that built the model also evaluated whether it works and uses it inside its health system. However, the level of validation varies across institutions, and the tools are not available outside their respective hospital systems.”

Gottlieb cited survey findings showing that 65% of U.S. hospitals use AI models, but only 61% of hospitals using predictive models test most or all of them for accuracy against local data, while 44% evaluate them for bias.

Gottlieb proposed that Congress authorize FDA to certify healthcare institutions to develop and deploy specified categories of clinical AI under standardized requirements for validation, monitoring, algorithm changes and performance reporting.

Under the proposed framework, lower-risk models could initially be deployed based on evidence of analytical validity, with clinical utility established through subsequent real-world use. Higher-risk applications could require premarket clinical evaluation.

Fully autonomous systems, patient-facing diagnostic tools and models distributed outside certified institutions would remain subject to conventional FDA medical device requirements, he said.

Container Closure System Guidance Comments

Pharmaceutical Research and Manufacturers of America (PhRMA) says it appreciates FDA’s attention to container closure systems in light of the significant technological advances that have emerged in material science and drug delivery systems since the agency’s 1999 guidance on the same topic. In a comment letter on an

FDA Launches Rare Disease Framework Initiative

FDA says it is launching a Rare Disease Framework Initiative “to reimagine the development of drugs and biologics for rare diseases.” In a statement, which includes a link to a new white paper on the initiative, the program will “develop an optimized, fit-for-purpose framework that reflects the realities

Bioresearch Monitoring Issue at Lovelace Institute

A 4/7-4/18/2025 FDA Bioresearch Monitoring Program inspection at Lovelace Biomedical Research Institute in Albuquerque, NM, found objectionable conditions in the facility’s role in conducting two nonclinical laboratory studies. A 10/6 Warning Letter says the specific violation was: * failure of the quality assurance unit to

CGMP Violations at Silgan Dispensing Systems Unit

A 3/26-4/2 FDA inspection of the Silgan Unicep Packaging drug manufacturing facility in Spokane, WA, found significant violations of current good manufacturing practice (CGMP) regulations for finished drugs. A 10/5 Warning Letter says the specific violations were: * failure to thoroughly investigate any unexplained discrepancy or failure

Read more

Container Closure System Guidance Comments

FDA Launches Rare Disease Framework Initiative

Bioresearch Monitoring Issue at Lovelace Institute

CGMP Violations at Silgan Dispensing Systems Unit