DL4papers: a deep learning approach for the automatic interpretation of scientific articles

Area of Science:

• Bioinformatics
• Computational Biology
• Precision Medicine

Background:

• The increasing volume of scientific literature presents challenges in extracting actionable insights for precision medicine.
• Manual curation of research papers to identify relationships between genes, drugs, and diseases is time-consuming and inefficient.

Purpose of the Study:

• To develop an automated method for extracting relevant relationships between keywords from scientific documents.
• To facilitate the identification of novel treatments and prediction of drug responses in precision medicine.

Main Methods:

• A deep learning-based method, DL4papers, was developed to analyze and interpret scientific papers.
• The method takes user-defined keywords as input and returns a ranked list of relevant documents.
• DL4papers highlights specific text fragments within documents that contain the queried associations.

Main Results:

• DL4papers demonstrated superior performance compared to existing methods on a cancer-related corpus.
• The method achieved high reliability, with the top two retrieved documents consistently containing relevant keyword associations.
• The system effectively identifies and highlights relevant text segments, reducing the need for full paper review.

Conclusions:

• DL4papers offers an accurate and efficient tool for researchers in precision medicine.
• The method accelerates the discovery of relationships crucial for advancing personalized treatment strategies.
• Automated text mining of scientific literature is essential for the progress of precision medicine.