Automatic Extraction of Medication Names in Tweets as Named Entity Recognition

TL;DR

This paper presents a method for recognizing medication mentions in tweets using ensemble fine-tuning of BERT models, achieving high accuracy in identifying medical references in social media posts.

Contribution

It introduces an ensemble approach with Megatron-BERT models for medication named entity recognition in tweets, improving upon previous methods.

Findings

Achieved a strict F1 score of 0.764 on test data.

Ensemble of five Megatron-BERT models outperforms individual models.

Demonstrated effectiveness of BERT-based models for social media medical text analysis.

Abstract

Social media posts contain potentially valuable information about medical conditions and health-related behavior. Biocreative VII Task 3 focuses on mining this information by recognizing mentions of medications and dietary supplements in tweets. We approach this task by fine tuning multiple BERT-style language models to perform token-level classification, and combining them into ensembles to generate final predictions. Our best system consists of five Megatron-BERT-345M models and achieves a strict F1 score of 0.764 on unseen test data.