Lucie M. Gattepaille's research while affiliated with Uppsala Monitoring Centre and other places

... Most national centres review case reports before they are sent to Vig-iBase, however, it is important to note that the information in Vigi-Base comes from a variety of sources, and the likelihood that the suspected ADR is drug-related is not the same in all cases. Once sent to VigiBase, ICSRs can be viewed in VigiLyze, a web-based advanced analytical tool developed by UMC that can be used to browse Vigi-Base data and to support signal detection (Van Hunsel, Gattepaille, et al., 2022). ...

... There clearly remains much confusion and lack of clarity around the scope of ML and AI and the usage as discussed in a systematic review [8] shows there is clearly a huge increase in published research on AI-based ML [7]. This is further illustrated by the wide range of examples of original research covering applications as diverse as predicting drug approvals [9], automated patient-reported adverse event [10] and drug coding [11], and adverse event report causality assessment [12], and disease prediction [13] and its role in supporting decision making by safety experts during signal validation [14]. This issue also contains perspectives from different stakeholders and data networks [15][16][17], insights and challenges into how ML can help facilitate identifying the completely unexpected 'black swan events' [18] and insights into how ML is making inroads into causal inference and telehealth and in resource-limited settings [19][20][21]. ...

... The MCDs will be extended, where available, with information extracted from, and tracked across, clinical narratives using NLP-contextualised language models such as BERT. This builds upon existing healthcare NLP applications and annotated datasets, such as WEB-RADR (extracting events related to adverse drug reactions) 13 and AVERT (mining mental health narratives from clinical letters). 14 The data include over four billion annotations over 12 years in a large mental health and community provider trust, plus inputs from other regions. ...

... There are other inherent limitations of the passive pharmacovigilance systems, such as a lack of denominators and underreporting, which have been described elsewhere [34]. On the other hand, the capacity to identify potential disproportionality signals is also increased in a global pharmacovigilance database over smaller national databases [35]. Moreover, the possibility of duplicate reporting cannot be ruled out; nevertheless, we used a de-duplicated dataset version to minimize this potential bias. ...

... Notably, the WEB-RADR and WEB-RADR 2 projects 1 explored the use of social media for drug safety monitoring scenarios and have published recommendations for the integration of social media for PV purposes [13] and guidance regarding related data collection good practices [14]. Furthermore, in the context of these projects Adverse Events (AE) recognition tasks were elaborated [15] and an evaluation/benchmarking dataset was published to facilitate the evaluation and the comparison of relevant algorithms and ICT tools [16]. ...

... The semi-automated pipeline published by [14] supports extracting ADR pairs from adverse events databases using statistical BPCNN algorithm for Natural Language Processing. Among other classical approaches commonly used in NLP, distributional semantics based on patterns of ADR co-reporting [15], Hidden Markov Models [16] or disproportionality analysis (DPA) [17] were already attempted to perform ADR detection. In 2012, Gurulingappa, Harsha et al. published an open-source reference dataset and developed a dictionary-based algorithm for extraction of adverse drug events in PubMed literature [18,19]. ...