Pharmaceutical Medicine, Год журнала: 2025, Номер unknown

Опубликована: Апрель 21, 2025

Pharmacovigilance is the science of collection, detection, and assessment adverse events associated with pharmaceutical products for ongoing monitoring understanding those products' safety profiles. Part this process, signal management, encompasses activities validation/confirmation, evaluation, ultimately, final as to whether a constitutes new causal drug reaction. Artificial intelligence group technologies including machine learning natural language processing that are revolutionizing multiple industries through intelligent automation. Here, we present critical evaluation studies leveraging artificial in management characterize benefits limitations technology, level transparency, our perspective on best practices future. To end, PubMed Embase were searched cumulatively terms pertaining intelligence, learning, or processing. Information model used, hyperparameter settings, training/testing data, performance, feature analysis, more was extracted from included articles. Common detection methods k-means, random forest, gradient boosting machine. Machine algorithms generally outperformed traditional frequentist Bayesian measures disproportionality per various metrics, showing potential utility advanced detection. In validation typically applied. Overall, methodological transparency mixed only some leveraged "gold standard" publicly available positive negative control datasets. innovation pharmacovigilance being driven by models, particularly part because high-performing bagging such forest These may be well poised accelerate progress field when used transparently ethically. Future research needed assess applicability these techniques across therapeutic areas classes broader industry.

Язык: Английский