Table 3.

Publicly available FDA documents for promoting AI-powered LMs in regulatory applications.

Data sets	Descriptions	Potential use in LMs	URL of data files
Drug labeling	Drug labeling comprises a summary of information for safe and effective use of the drug, which is proposed by manufacturer and approved by FDA	Drug labeling could be a useful resource (>120 000 product labelings) to develop biomedical named-entity recognition/normalization, and relation extraction between drug and AEs, drug–drug interaction, etc.	https://dailymed.nlm.nih.gov/dailymed/spl-resources-all-drug-labels.cfm
FAERS	FAERS is a database that contains information on AE and medication error reports submitted to FDA	FAERS is designed to support the post-marketing safety surveillance program for drug and therapeutic biologic products of the FDA. There are more than 19 million case reports in FAERS; AI-powered LMs could be applied to carry out AE detection, causal relationship extraction, etc.	https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html
Orange Book	Orange Book identifies drug products approved on basis of safety and effectiveness by FDA under the Federal Food, Drug, and Cosmetic Act and related patent and exclusivity information	Orange book provides crucial regulatory information, such as biological equivalence, reference listed drug (RLD), Reference Standard (RS), and patent status. This information could be included in AI-powered LMs to compare drug product information with RLD and RS to facilitate abbreviated new drug application (ANDA) submissions	www.fda.gov/drugs/drug-approvals-and-databases/orange-book-data-files
Drugs@FDA	Drugs@FDA includes most drug products approved since 1939. Most patient information, labels, approval letters, reviews, and other information are available for drug products approved since 1998	Drugs@FDA provides rich information on drug approval history, which could be used as AI-powered LMs to explore underlying reasons for labeling changes and increase business success	www.fda.gov/drugs/drug-approvals-and-databases/drugsfda-data-files
FDA Guidance Documents	Guidance documents describe FDA’s interpretation of policy on a regulatory issue (21 CFR 10.115(b)). These documents usually discuss more specific products or issues that relate to design, production, labeling, promotion, manufacturing, and testing of regulated products	FDA Guidance Documents could be useful to implement AI-powered LMs for standardizing and monitoring crucial steps in drug discovery and development in terms of their consistency and alignment with regulatory requirements	www.fda.gov/regulatory-information/search-fda-guidance-documents
FDA Acronyms and Abbreviations	FDA Acronyms and Abbreviations database provides a quick reference to acronyms and abbreviations related to FDA activities	Emphasis of FDA Acronyms and Abbreviations is on scientific, regulatory, government agency, and computer application terms. The database includes some FDA organizational and program acronyms. It is a useful resource to define vocabularies in AI-powered LMs and increase model generalization	www.accessdata.fda.gov/scripts/cder/acronyms/index.cfm