This week NEJM weighs in on two hot topics, the impact of the CHEVRON Supreme Court decision on health care policy & law, and an article on the ethics of AI in healthcare.
Sachs RE & Brown ECF (2024) Supreme Power - The loss of judicial defence to health agencies. Online first. https://www.nejm.org/doi/full/10.1056/NEJMp2408197
Sim I & Cassel C (2024) The Ethics of Relationa AI - expanding and implementing the Belmont principles. NEJM 391:193-6. https://www.nejm.org/doi/full/10.1056/NEJMp2314771
(I'm a subscriber, so I'm not sure if either or both are open access.)
The two articles lack abstracts, so I've included an AI Summary below.
###
AI CORNER
[Start Chat GPT 4]
Supreme Power — The Loss of Judicial Deference to Health Agencies
New England Journal of Medicine, July 17, 2024
The U.S. Supreme Court's recent ruling in Loper Bright Enterprises v. Raimondo marks a significant shift in judicial authority, ending the Chevron doctrine, which had guided courts to defer to agency interpretations of ambiguous statutes for 40 years. This decision reallocates power from Congress and administrative agencies to the judiciary, posing potential challenges for health agencies like CMS and FDA.
Key implications include:
- Increased litigation and regulatory uncertainty for the healthcare industry.
- Diminished expert authority over health regulations, risking patient safety and public health.
- Potential disruptions in Medicare payments and challenges to FDA's regulatory capabilities.
Justice Elena Kagan's dissent highlights the loss of valuable agency expertise in statutory interpretation. The decision's lack of clear guidelines for lower courts could lead to more legal challenges and uncertainty, particularly impacting technical healthcare regulations.
The ruling is part of a broader trend of the Court enhancing judicial power, potentially favoring well-funded industry litigants over government regulations, with significant consequences for public health and safety.
[BQ - Post-Chevron, a court just found that HHS was wrong to find that "sexual discrimination" included the concept of "transgender," doing so, literally, by using 1972 dictionaries. That is, it was a "legal" or "judicial" finding (using the judicial method of dictionaries) not a "policy" finding by HHS. Tennessee v Becerra, July 2024. Here.] [See my appended note at the bottom of Tenn. v Becerra blog, about dictionaries and the definition of "necessary."]
###
The Ethics of Relational AI — Expanding and Implementing the Belmont Principles
New England Journal of Medicine, July 18, 2024
The rise of relational AI, exemplified by ChatGPT, marks a new era in artificial intelligence, transforming how AI interacts with humans through natural language. This shift brings additional ethical challenges beyond those associated with traditional predictive AI.
Key points include:
- Ethical Principles: The Belmont Report's principles of beneficence, respect for persons, and justice must be expanded to include clinicians as well as patients. These principles are crucial for guiding the ethical implementation of relational AI in healthcare.
- Applications and Risks: Relational AI can draft patient messages, provide treatment information, and even create lifelike avatars of clinicians. However, these advancements come with risks such as AI "hallucinations," biased training data, and potential misuse.
- [BQ: For new work by Scott Gottlieb on AI matching clinicians in a "treatment setting," here.]
- Transparency and Fairness: Ensuring AI transparency about data use and decision-making processes is vital. Fairness must be addressed by using representative data sets and managing biases. The benefits of AI should be equitably distributed, considering both commercial gains and efficiency improvements.
- Call for Action: A national network of health AI ethics centers is proposed to develop guidelines, monitor implementation, and ensure AI adheres to ethical principles. These centers should involve diverse expertise and active community participation.
The integration of relational AI in healthcare necessitates a robust ethical framework to protect patients, clinicians, and society at large.
[END Chat GPT 4]
