Vol. 6, Issue 1, Part B (2024)

Introduction: Artificial intelligence, particularly through machine learning, leverages algorithms and past experiences to make predictions. Lately, there's been a growing interest in incorporating more AI into the pharmacovigilance of products already on the market, as well as those pharmaceuticals still in development. Objective: The goal of this study was to pinpoint and explain how artificial intelligence is being utilized in pharmacovigilance by conducting a thorough review of existing literature. Methods: Searches were conducted in the Embase and MEDLINE databases for articles published between January 1, 2019, and July 9, 2024. Search terms like ‘pharmacovigilance,’ ‘patient safety,’ ‘artificial intelligence,’ and ‘machine learning’ were used specifically in the title or abstract. Scientific articles were reviewed and synthesized that discussed the application of AI across various aspects of patient safety or pharmacovigilance, following a pre-defined data extraction template. Articles that had incomplete information, as well as letters to the editor, notes, and commentaries were excluded. Results: A total of 66 articles were reviewed for this evaluation. The most pertinent studies on artificial intelligence primarily concentrated on machine learning, particularly its application in enhancing patient safety. This included identifying adverse drug events (ADEs) and adverse drug reactions (ADRs) (57.6%), processing safety reports (21.2%), extracting drug-drug interactions (7.6%), pinpointing populations at high risk for drug toxicity or providing guidance for personalized care (7.6%), predicting side effects (3.0%), simulating clinical trials (1.5%), and incorporating prediction uncertainties into diagnostic classifiers to boost patient safety (1.5%). AI has played a role in detecting safety signals through automated processes and training with machine learning models. However, it's important to note that the findings may not be universally applicable, as the data types varied across the different sources. Conclusion: Artificial intelligence is a game-changer when it comes to processing and analyzing vast amounts of data, especially in the context of various diseases. With the help of automation and machine learning models, we can streamline pharmacovigilance processes, making it easier to shift through safety-related information. However, we still need more research to determine whether these optimizations truly enhance the quality of safety analyses. Looking ahead, we can expect its usage to grow, particularly in predicting side effects and adverse drug reactions (ADRs).