Unlock the algorithms: regulation of adaptive algorithms in reproduction

In the USA, the Food and Drug Administration plans to regulate artificial intelligence and machine learning software systems as medical devices to improve the quality, consistency, and transparency of their performance across specific age, racial, and ethnic groups. Embryology procedures do not fall...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Fertility and sterility Ročník 120; číslo 1; s. 38
Hlavní autor: Curchoe, Carol Lynn
Médium: Journal Article
Jazyk:angličtina
Vydáno: United States 01.07.2023
Témata:
Algorithms
Artificial Intelligence
Humans
Machine Learning
Reproduction
Software
regulatory affairs
FDA
Artificial intelligence
medical AI
ISSN:1556-5653, 1556-5653
On-line přístup:Zjistit podrobnosti o přístupu
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In the USA, the Food and Drug Administration plans to regulate artificial intelligence and machine learning software systems as medical devices to improve the quality, consistency, and transparency of their performance across specific age, racial, and ethnic groups. Embryology procedures do not fall under the federal regulation of "CLIA 88." They are not tests per se; they are cell-based procedures. Likewise, many add-on procedures related to embryology, such as preimplantation genetic testing, are considered "laboratory-developed tests" and are not subject to Food and Drug Administration regulation at present. Should predictive artificial intelligence algorithms in reproduction be considered medical devices or laboratory-developed tests? Certain indications certainly carry a higher risk, such as medication dosage, where the consequences of mismanagement could be severe, whereas others, such as embryo selection, are noninterventional (selecting from a patient's own embryos and the course of treatment does not change) and present little to no risk. The regulatory landscape is complex, involving data diversity and performance, real-world evidence, cybersecurity, and postmarket surveillance.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
ObjectType-Review-3
content type line 23
ISSN:1556-5653
1556-5653
DOI:10.1016/j.fertnstert.2023.05.152