A breakthrough in pharmaceutical safety surveillance is reshaping how healthcare systems detect adverse drug events. Researchers have developed a large language model capable of automatically identifying safety signals from clinical notes embedded within electronic health records, according to a study published in Drug Safety (2026). This technological advance addresses a critical limitation of current pharmacovigilance systems, which rely heavily on labour-intensive manual review processes that often fail to capture adverse events buried within unstructured clinical documentation. The AI model demonstrates high accuracy in extracting adverse drug event information, achieving 95% accuracy in clinical note processing and 88% in event identification. By automating detection of safety signals across vast healthcare datasets, this approach could enable earlier identification of emerging drug safety concerns, ultimately strengthening post-marketing surveillance and protecting patient safety at scale. Read the full article on GMJ Newsroom.
Was this article helpful?
