Applying machine learning to pharmacovigilance data: A proof‐of‐concept study
Romain Barus, Pauline Schiro, Jean‐Luc Faillie, Romane Freppel, Johana Béné, Claire de Canecaude, François Montastruc · 2026
Read Applying machine learning to pharmacovigilance data: A proof‐of‐concept study by Romain Barus, Pauline Schiro, Jean‐Luc Faillie, Romane Freppel, Johana Béné, Claire de Canecaude, François Montastruc (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13619036.txt.
Description
Romain Barus and six colleagues offer a 2026 proof of concept for machine learning in drug-safety surveillance, a field dominated by manual review. From the French national pharmacovigilance database they pulled every report involving warfarin, the blood thinner notorious for gastrointestinal bleeding. They trained an extreme gradient boosting model to sort reports into gut-bleeding versus other adverse reactions. An explanation method then exposed which features drove its calls.
After cleaning, one thousand forty-five reports remained. On a test set the model reached moderate discrimination, catching eighty percent of true bleeding reports with overall accuracy just under seventy percent. The explanation analysis ranked age, baseline gastrointestinal bleeding risk, antiplatelet use, and a history of kidney dysfunction as the top features, a list pharmacologists checked afterward. The authors present the pairing of model plus explanation as a template for triaging safety reports and surfacing risk factors.
Catalog record
| CREATOR | Romain Barus, Pauline Schiro, Jean‐Luc Faillie, Romane Freppel, Johana Béné, Claire de Canecaude, François Montastruc |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29190 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |