@article{fdi:010086821,
  title = {{L}i{SA} : an assisted literature search pipeline for detecting serious adverse drug events with deep learning},
  author = {{M}artenot, {V}. and {M}asdeu, {V}. and {C}upe, {J}. and {G}ehin, {F}. and {B}lanchon, {M}. and {D}auriat, {J}. and {H}orst, {A}. and {R}enaudin, {M}. and {G}irard, {P}. and {Z}ucker, {J}ean-{D}aniel},
  editor = {},
  language = {{ENG}},
  abstract = {{I}ntroduction: {D}etecting safety signals attributed to a drug in scientific literature is a fundamental issue in pharmacovigilance. {T}he constant increase in the volume of publications requires the automation of this tedious task, in order to find and extract relevant articles from the pack. {T}his task is critical, as serious {A}dverse {D}rug {R}eactions ({ADR}s) still account for a large number of hospital admissions each year. {O}bjectives: {T}he aim of this study is to develop an augmented intelligence methodology for automatically identifying relevant publications mentioning an established link between a {D}rug and a {S}erious {A}dverse {E}vent, according to the {E}uropean {M}edicines {A}gency ({EMA}) definition of seriousness.{M}ethods: {T}he proposed pipeline, called {L}i{SA} (for {L}iterature {S}earch {A}pplication), is based on three independent deep learning models supporting a precise detection of safety signals in the biomedical literature. {B}y combining a {B}idirectional {E}ncoder {R}epresentations from {T}ransformers ({BERT}) algorithms and a modular architecture, the pipeline achieves a precision of 0.81 and a recall of 0.89 at sentences level in articles extracted from {P}ub{M}ed (either abstract or full-text). {W}e also measured that by using {L}i{SA}, a medical reviewer increases by a factor of 2.5 the number of relevant documents it can collect and evaluate compared to a simple keyword search. {I}n the interest of re-usability, emphasis was placed on building a modular pipeline allowing the insertion of other {NLP} modules to enrich the results provided by the system, and extend it to other use cases. {I}n addition, a lightweight visualization tool was developed to analyze and monitor safety signal results.{C}onclusions: {O}verall, the generic pipeline and the visualization tool proposed in this article allows for efficient and accurate monitoring of serious adverse drug reactions from the literature and can easily be adapted to similar pharmacovigilance use cases. {T}o facilitate reproducibility and benefit other research studies, we also shared a first benchmark dataset for {S}erious {A}dverse {D}rug {E}vents detection.},
  keywords = {{A}dverse drug events ; {A}ssisted literature review ; {D}eep {L}earning ; {NLP}},
  booktitle = {},
  journal = {{BMC} {M}edical {I}nformatics and {D}ecision {M}aking},
  volume = {22},
  numero = {1},
  pages = {[16 p.]},
  year = {2022},
  DOI = {10.1186/s12911-022-02085-0},
  URL = {https://www.documentation.ird.fr/hor/fdi:010086821},
}