Dhruvilkumar Ashokbhai Chodvadiya / Earth & Environmental Sciences / Faculty Mentor: Yike Shen
The interdisciplinary field of environmental health sciences is rapidly evolving. Although knowledge graphs summarize cohort study findings, publication searches remain manual. An interactive webtool linking graph edges to publications can increase efficiency. Additionally, while machine learning (ML) models are widely applied to predict chemical toxicity, many users lack the Python expertise to develop, train, and validate these models. A web-based ML tool providing direct predictions will improve accessibility. We aim to develop webtools for knowledge sharing and toxicity prediction in environmental health. We built two interactive web tools using VS Code and React, integrated via GitHub, and deployed through Vercel with GoDaddy domains. The knowledge graph tool, developed with Cytoscape and web-scraping, links edges to publications through DOIs. The ML tool leverages AWS (API Gateway, Lambda, S3) to process user-uploaded data, apply a pre-trained model, and return ecotoxicity predictions via API Gateway. Both MIT-licensed websites include informative landing pages with background information. We launched ecotoxicity.org for ML-based toxicity prediction and cohortnetwork.org, an interactive multi-layer knowledge graph linking exposures, outcomes, and associations in environmental health cohorts. Our developed webtools provide a valuable resource for the environmental health science community for more efficient knowledge sharing and more accurate toxicity prediction.
Leave a Reply