
PathwayPulse is an autonomous biotech arbitrage engine that surfaces cross-indication drug repurposing signals before they reach press releases or stock prices. The earliest evidence of a repurposing thesis rarely appears in one place. It is scattered across bioRxiv and medRxiv preprints, Reddit biotech communities, ClinicalTrials.gov registries, and conference abstract portals which occur months before peer-reviewed publication. No single analyst can read all of that in real time. PathwayPulse can. Our ingestion swarm pulls from these sources in parallel. Bright Data's Scraping Browser is the critical layer for JavaScript-heavy targets that resist standard scrapers, including ACR rheumatology conference abstracts and ClinicalTrials.gov's Angular SPA, providing bot evasion and reliable rendering where free APIs fall short. A Catalyst Calendar enriches discovered trial IDs with structured readout dates from the ClinicalTrials.gov API v2, flagging overdue and imminent data events. Every record runs through a dual-model AI pipeline. DeepSeek-V3 via AI/ML API triages hundreds of documents concurrently, extracting structured cross-pollination events such as what pathway, from which disease, into which new indication, with what confidence. GPT-4o then evaluates immunological soundness and renders a Bear / Bull / Neutral intelligence report with a ranked watch list. Results appear in a live Streamlit dashboard: an interactive arbitrage matrix node graph with confidence-weighted edges, a written intelligence report, and a color-coded trial readout calendar. Enter a pathway like IL-6 signaling or JAK-STAT and get a ranked set of repurposing theses in minutes.
31 May 2026