An agentic system that takes a patient's chief complaint, walks through structured follow-up questions, retrieves the relevant clinical guidelines, and produces a draft History and Physical note for the attending to refine.
Claude APIModel Context Protocol (MCP)Pinecone (vector DB)StreamlitVercel deploy
What it does
Ingests a chief complaint and patient demographics in plain language
Runs a structured pre-visit interview, branching on symptom acuity
Retrieves relevant clinical guidelines via RAG over a curated medical knowledge base
Drafts a structured History and Physical with citations and disclaimers
Surfaces red-flag findings the attending should review first
Logs the full reasoning trace for audit and follow-up
Co-Teaching Faculty
Medical resident archetype (sample profile)
Practicing resident · Northwestern Feinberg
Our pre-med track is co-taught by a working medical resident — recruited from Northwestern Feinberg, Loyola Stritch, or another leading academic medical center — alongside a senior software engineer. Your child works directly with a physician-in-training who has shipped clinical workflows in real hospital settings, and who can speak to where AI fits ethically and safely into patient care today. The mentor relationship continues past the program: pre-med students who perform well receive a personal letter of recommendation and direct introductions into the broader practitioner network.
“Students leave understanding what an attending actually wants from a draft note — and what AI should never decide alone.”
Curriculum Deep-Dive
Week 2 — Clinical AI deep-dive
The pre-med cohort splits off from the shared Week 1 foundations and works directly with the medical resident on the data, workflows, and ethical guardrails unique to clinical AI.
01De-identified patient intake datasets and HIPAA-aware prompt engineering
02Clinical guideline retrieval — chunking medical literature for high-precision RAG
03Structured H&P drafting with citation grounding and disclaimer scaffolding
04Red-flag detection and acuity classification for triage workflows
05Hands-on architecture review with the resident: where AI fits and where it does not
06Capstone scoping: pick a real clinical workflow, write the spec, agree on an evaluation rubric
Week 1 foundations are shared across all 5 tracksWeek 3 builds + admissions coaching
Sample student work
What your child could build.
Sample capstone projects illustrating the scope of Pre-Med work. Inaugural cohort is Summer 2028 — these are not real-alumni outcomes.
Sample student work — Aiden L. (rising 11th)
Pre-visit triage agent for primary care
Built an agent that takes a patient's chief complaint, asks five branching follow-up questions, and produces a draft note flagging acute findings for the attending. Designed against an open clinical-guidelines corpus — never against protected health information.
Claude + MCP + Pinecone + Streamlit
Sample student work — Priya M. (rising 12th)
Specialty referral routing agent
Built an agent that reviews a primary-care visit summary and recommends a specialty referral with rationale, citing the guidelines that drove the routing decision. Tested across 50 synthetic patient histories with a human-reviewed evaluation rubric.
Claude + LangGraph + pgvector + Next.js
Why this track wins for admissions
A capstone an admissions reader can verify.
An admissions reader at a top medical-track university — BS/MD, BS/DDS, or biology with a clinical-research focus — is looking for the rare applicant who can articulate, in their own words, what the practice of medicine actually requires of an AI system. Most strong pre-med applicants present community service, lab experience, and shadowing — all important, all common. Far fewer can sit across from a physician interviewer and discuss the design tradeoffs in patient-intake automation, the ethics of model uncertainty in triage, or what a working H&P workflow looks like inside an electronic health record. Your child finishes the program with a working clinical-intake agent at a personal URL, an architecture writeup an admissions reader can verify, and a Common App essay built around the project — a piece of writing that makes their interest in medicine specific, technical, and verifiably their own.