Masters Thesis: COVID-LEAP
Azure-hosted knowledge base with BERT semantic search across 4.7M coronavirus research paragraphs and 500K clinical trials.
COVID-LEAP (Literature Exploration Analysis Platform)
I wanted to be part of the COVID-19 fight. To use my skills and time for something good. I set two objectives for the project:
- Support Pharma/BioTech decision-making with timely and highly visual summarisation of the COVID-19 vaccine development landscape.
- Support time-pressured researchers with more effective information retrieval. Identify the most relevant papers from the highly dynamic COVID-19 academic literature to answer their information needs.
I architected and developed an Azure-cloud hosted research knowledge base for analysis and search, using high-volume COVID-19 papers (400K) and clinical trials info (500K) for evaluation.
Vaccine trials information from multiple sources is automatically aggregated and can be explored in a highly visual way to help track and understand the dynamic vaccine development landscape.
State-of-the-art search with BERT-based strategies locates highly relevant academic papers to a biomedical question. The best strategy outperforms the "go to" PubMedCentral search engine as evaluated by a medical expert.
Artifacts
- Master's Degree Thesis Report (COVID-LEAP) - grade 90%