AI-based Approach for Safety Signals Detection from Social Networks: Application to the Levothyrox Scandal in 2017 on Doctissimo Forum

Valentin Roche, Jean-Philippe Robert, Hanan Salam

Research output: Contribution to journalArticlepeer-review


Social media can be an important source of information facilitating the detection of new safety signals in pharmacovigilance. Various approaches have investigated the analysis of social media data using Artificial Intelligence such as Natural Language Processing (NLP) techniques for detecting adverse drug events. Existing approaches have focused on the extraction and identification of Adverse Drug Reactions (ADR), Drug-Drug Interactions (DDI) and drug misuse. However, non of the works tackled the detection of potential safety signals by taking into account the evolution in time of relevant indicators.
Moreover, despite the success of deep learning in various healthcare applications, it was not explored for this task.
We propose an Artificial Intelligence (AI) based approach for the detection of potential pharmaceutical safety signals from patients' reviews that can be used as part of the pharmacovigilance surveillance process to flag the necessity of immediate attention and encourage the conduct of an in-depth pharmacovigilance investigation.
We focus on the Levothyrox case in France which triggered a huge attention from the media following the change of the medication formula, leading to an increase in the frequency of adverse drug reactions normally reported by patients.
Our approach is two-fold. (1) We investigate various NLP-based indicators extracted from patients' reviews including words and n-grams frequency, semantic similarity, Adverse Drug Reactions mentions, and sentiment analysis.
(2) We propose a deep learning architecture, named Word Cloud Convolutional Neural Network (WC-CNN) which trains a CNN on word clouds extracted from the patients comments. We study the effect of different time resolutions (day, week, month, combined) as well as different NLP pre-processing techniques on the performance of the model.
Our results show that the proposed indicators could be used in the future to effectively detect new safety signals. Moreover, the deep learning model trained on word clouds extracted at monthly resolution outperform models trained on data extracted at daily, weekly, and combined time resolutions with an accuracy of 75%.
Original languageEnglish (US)
JournalArtificial Intelligence in Medicine
StateSubmitted - 2021


Dive into the research topics of 'AI-based Approach for Safety Signals Detection from Social Networks: Application to the Levothyrox Scandal in 2017 on Doctissimo Forum'. Together they form a unique fingerprint.

Cite this