DL4papers

DL4papers applies deep learning to extract and rank associations between biomedical entities from scientific literature to accelerate precision medicine analyses of data from next-generation sequencing and preclinical studies.

Key Features:

• Automated Relation Extraction: Identifies and extracts relevant associations between specific keywords provided as input queries from scientific articles.
• Ranked Output: Returns a ranked list of scientific papers containing significant associations among the specified keywords, prioritizing the most relevant connections.
• High Reliability in Top Results: Empirical evaluations report 100% accuracy for the first two results on average in retrieving highly relevant documents.
• Highlighting Relevant Text Fragments: Highlights specific text fragments within each document where the associations between input keywords occur.
• Deep Learning Modeling: Employs a deep learning–based model trained to discern patterns and associations between keywords and outperforms traditional text-mining methods in cancer-related corpora.

Scientific Applications:

• Precision Medicine: Rapidly identify relationships between genes and their mutations, and between drug responses and treatments for specific diseases.
• Next-Generation Sequencing and Preclinical Study Curation: Facilitate sifting and synthesis of literature generated by next-generation sequencing and novel preclinical studies.
• Cancer Research: Extract and prioritize cancer-related associations from literature, supported by superior performance on cancer-related corpora.
• Drug Response Analysis: Support identification of literature evidence relevant to predicting drug responses and treatment outcomes.

Methodology:

DL4papers uses a deep learning–based approach that processes textual data from scientific literature and is trained to discern patterns and associations between keywords within complex documents, with performance evaluated against traditional text-mining methods on cancer-related corpora.