• 

  Synapse, Probability of Trial Success Tracker

• 

Synapse

Real-time AI-powered clinical trial portfolio intelligence

Inspiration

Tracking clinical trial outcomes is critical for biotech investors and pharma strategy teams, but it's tedious. Analysts spend hours scanning headlines, cross-referencing FDA announcements, and manually updating probability scores. We wanted to automate the grunt work so teams can focus on decisions, not data gathering.

What it does

Synapse provides real-time monitoring for your clinical trial portfolio, automatically scanning news for FDA decisions, safety signals, and competitor events. It uses Claude to analyze findings, update probability of success scores, and explain what changed—giving you instant visibility into clinical catalysts.

How we built it

We integrated Yutori for web browsing and news retrieval, Claude Opus for analysis and synthesis, and built a clean interface to display portfolio changes with explanations. The system chains together search, analysis, and scoring into a continuous monitoring workflow. We also used Kiro & Claude Code.

Challenges we ran into

Parsing unstructured news into structured trial impacts was tricky. We also had to balance thoroughness (checking many sources) with speed (users need real-time results).

Accomplishments that we're proud of

The "explain why" feature—every score change comes with a plain-English rationale. It turns a black-box update into something teams can actually act on.

What we learned

LLMs are powerful for synthesis, but prompt design matters enormously for consistent, reliable outputs in domain-specific contexts like clinical trials.

What's next for Synapse

• Slack/email alerts for material changes
• Deeper integration with clinical trial databases (ClinicalTrials.gov)
• Confidence intervals on probability estimates

Built With

• claude-opus-4.5-(anthropic-api)
• javascript
• kiro
• react
• tailwind-css
• vite
• yutori-browsing-api