A doctor built a ventilator tool in 4 minutes using Claude
A peer-reviewed technical note published in Anaesthesia Critical Care & Pain Medicine documents something that should interest anyone experimenting with AI-assisted tooling: a single physician built a functional, tested clinical decision-support application in 4 minutes and 25 seconds, with no programming experience required.
What Was Built
The tool is called VentiProtect ARDS. It assists with real-time ventilator titration for patients with acute respiratory distress syndrome (ARDS), a condition where lung-protective ventilation is frequently under-applied at the bedside because integrated decision-support tools don’t exist.
The application computes six clinical values: predicted body weight, tidal volume, driving pressure, respiratory system compliance, mechanical power, and oxygenation index. It adds real-time color-coded alerts and an active suggestion engine that calculates exact corrective ventilator settings.
How It Was Built
The physician, Sacha Rozencwajg, used a structured eight-block prompt submitted to Claude Opus 4.6 via the Antigravity agentic development environment. The approach described in the paper is what the authors call physician prompt engineering: translating clinical knowledge into a natural-language specification that constrains LLM code generation, rather than writing code directly.
The prompt went through an iterative pre-prompt reasoning process before submission. The entire session ran 4 minutes 25 seconds.
Verification
Formula accuracy was tested using an independent automated test suite of 100 scenarios covering 1,400 individual checks. No computational errors were identified. The tool runs locally with no installation, no network dependency, and no patient data transmission.
The Operator Angle
The paper frames this as a proof-of-concept for a broader claim: domain experts in any field may be able to build working, accurate tools for their specific workflows without depending on software vendors or engineers. The authors are careful to note that a single self-performed case cannot establish this across all physicians, but the result is a clean data point.
The line that closes the abstract is worth keeping: “The LLM generates syntax. The physician generates medicine.”
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
