Clinical Trial AI Platform

Lumen is an AI-powered clinical trial management platform I built to streamline the entire trial design and planning workflow. Starting from a blank form, users generate complete, regulator-ready clinical trial protocols in seconds, then run feasibility assessments, regulatory alignment analyses, contract reviews, and budget estimates — all from a single tool backed by Anthropic Claude.

In my Healthcare Economics & Data class at Sloan, I've been studying the pharmaceutical industry. Policy pressure, including the IRA's shortened exclusivity windows for small molecule drugs and executive action on Most Favored Nation pricing has made reducing drug development time and cost more urgent than ever.

However, clinical trials are deeply inefficient due to regulatory requirements, enrollment challenges, and fragmented coordination. The inefficiency starts before a trial even begins in the protocol development phase. For example, a single Phase III protocol requires weeks of work from various stakeholders (medical writers, regulatory experts, statisticians) and even after submission, costly amendments are common.

After doing research into products in the market, I built Lumen to compress the clinical trial process, starting with protocol development. More features to be added soon.

Lumen is a Next.js 14 + Supabase web app powered by Claude.

• Protocol Generator: Fill a structured intake form or upload your notes. AI streams a complete ICH E6(R3) compliant protocol across 10 sections. The protocol is editable inline and exportable as a PDF.
• Protocol Intelligence: Score (0-100) any protocol across Regulatory Readiness, Scientific Rigot, and Feasibility of amendment risk and recruitment difficulty, and check FDA/EMA alignment via a compliance checklist.
• Contract Management: Upload site agreements or CRO contracts; AI extracts key terms in parallel and flags non-standard clauses by severity.
• Budget Planner: Input study parameters and receive a line-item budget with low/high ranges across standard cost categories.
• RAG Assistants: Build document-trained chatbot assistants from protocols, SOPs, or regulatory submissions for a persistent Q&A with semantic search.
• Tech Stack: Next.js 14, TypeScript, Tailwind CSS, Supabase, Anthropic Claude, Voyage AI, jsPDF

Lumen demonstrates what an AI-native clinical operations platform can look like. It compresses weeks of specialist work into a guided, single-tool workflow.