A Comprehensive Review of Generative AI Applications in Drug Safety

Abstract

The forecast of drug-drug interactions (DDIs) and adverse drug reactions (ADRs) is something that has a direct relationship with the safety of the patient. Boom in polypharmacy is an additional justification for the above statement. The objective of the traditional models, such as clinical trials and post-market surveillance, is to identify the DDIs and ADRs. However, these models are limited mainly in the diversity, granularity, and timeliness of the data that they rely on. Generative AI has come on the plate as a possible option as it is capable of creating new drug combinations and thus predicting the likely perils by applying machine learning methodologies. This survey on the continued developments related to the use of generative AI in both DDI and ADR will cover GANs, VAEs, and specifically, transformers. The authors discuss the ethical and legal implications of using these models in clinical practice in addition to outlining data sources and methodological difficulties. Future directions to enhance generative AI in pharmacovigilance are also highlighted.

Ähnliche Arbeiten