Building a Knowledge Network of Drug-Drug Interactions Using Natural Language Processing and Graph Databases
Ruchi Jakhmola Mani, Harsh Lalwani, Angamba Meetei Potshangbam, Shikha Rani, Kriti Katare, Abhishek Chauhan · 2025
Drug-drug interactions (DDIs) are a critical healthcare concern, often causing adverse effects and poor treatment outcomes. While many studies address DDIs, integrating this knowledge in a structured, accessible format remains a challenge. This study proposes a data-driven method using Natural Language Processing (NLP) and graph networks to extract and analyze DDI patterns. PubMed articles related to a specific drug are mined using Python, and interacting compounds are extracted from abstracts via Named Entity Recognition (NER) and text mining. These mentions form a curated DDI dataset used to build a knowledge network illustrating co-prescribed drugs, studied interactions, and reported effects. The resulting graph enables visual and query-based exploration of known and potential DDIs. Graph analytics help identify frequent drug pairs, interaction trends, and research gaps. This platform aids biotech and pharma R&D professionals in visualizing past DDI studies and planning future research efficiently, emphasizing the value of automated literature mining.