A LightGBM classifier with 3,451 features from NLP, embeddings, and medical patterns detects dosing errors in 42,112 clinical narratives at 0.8725 test ROC-AUC.
clinical practice guidelines
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
BioBERT embeddings plus logistic regression reach ROC-AUC 0.794 for dosing-error prediction in clinical-trial text, with gradient boosting and similar models reaching 0.821-0.853.
citing papers explorer
-
Automated Detection of Dosing Errors in Clinical Trial Narratives: A Multi-Modal Feature Engineering Approach with LightGBM
A LightGBM classifier with 3,451 features from NLP, embeddings, and medical patterns detects dosing errors in 42,112 clinical narratives at 0.8725 test ROC-AUC.