Industries / Health / Medical devices
Medical sterilization
Medical sterilization kills the microbes on medical devices and supplies, in factories before they are sold and in hospitals between patients, mostly with steam, ethylene oxide gas or radiation, and proves by tests and records that each load came out sterile.
1891
Johnson & Johnson began industrial steam sterilization at its plant in New Brunswick, New Jersey, and by the next year was steam-sterilizing the sutures and surgical dressings it mass-produced, in a large sterilizer built under its scientific director, Fred Kilmer, by the company's account.
37
Milestones achieved across 6 domains.
9
To watch for: announced, not yet confirmed outside.
13
Open deliverables. Nobody has shipped them.
In one minute
Where it started
In 1879 Charles Chamberland, in Louis Pasteur's laboratory, designed the autoclave, a sealed vessel that heats its load with steam under pressure, and it soon spread to surgery departments. In 1891 Johnson & Johnson moved steam sterilization into a factory for the dressings and sutures it sold. Two methods that work without heat followed: ethylene oxide (EtO) gas, patented for the purpose in 1940, and radiation, first an electron beam on Ethicon's packed sutures in 1956 and then gamma rays from cobalt-60.
Where it stands
About half of all sterile devices in the US are sterilized with EtO, by the literature the FDA cites, and industry estimates put gamma rays at 30 to 40% of the world's single-use devices. A few firms and a few reactors carry the load: by evidence in a 2015 court case, STERIS and Sterigenics did about 85% of US contract sterilization, and about half the world's cobalt-60, the source of gamma rays, comes from power reactors in Ontario, by Nordion's account. Since the EPA raised its estimate of EtO's cancer risk about thirtyfold in 2016, plants have faced lawsuits and tighter air rules, which a 2025 presidential proclamation put off for about 40 of them.
What is open
Much of what the industry runs on is known only to the companies: which plant sterilizes which devices, how much cobalt-60 each reactor will deliver, and how much EtO each plant really releases. Most open jobs here turn scattered public records into tables anyone can check: FDA approval orders and registrations, air-monitoring results, hospital inspection reports and recall records. AI models that read long documents and quote their sources make this possible for a small team, and none of the jobs needs a sterilizer, a gas or a radiation source.
Domains
AchievedWatch forOpen
Domain
Progress
Last achieved
Watch for
Open
Ethylene oxide
This work sterilizes packed devices in sealed chambers with ethylene oxide (EtO), a gas that kills microbes without heat but is toxic, flammable and causes cancer, and then airs the gas back out of the products.
6 · 1 · 2
Last achievedFDA asks for less gas2019
Watch forChlorine dioxide in factories
2
openRadiation
Radiation sterilization sends gamma rays, electron beams or X-rays through sealed boxes of finished products, killing microbes without heat.
6 · 1 · 2
Last achievedX-ray lines multiply2026
Watch forX-ray takes gamma's work
2
openSteam and hospital sterile rooms
This is the work hospitals do to clean, pack and sterilize reusable surgical instruments between patients, mostly with steam, in a department of their own.
7 · 1 · 2
Last achievedQuick cycles tied to infections2024
Watch forCameras check trays at assembly
2
openProving it worked
This is the work of showing that a load really was sterilized: tests of the machine, strips and spore tests inside the load, and the records on which a load is released for use.
5 · 2 · 3
Last achievedStrips that pass failed cycles2020
Watch forSpore checks in seconds
3
openEmissions
This work is about the EtO that leaves sterilizing plants through vents, leaks and freshly sterilized goods, how much of it reaches neighbors, how risky that is, and the rules and lawsuits that follow.
7 · 2 · 2
Last achievedThe 2024 limits put off2025
Watch forExemptions tested in court
2
openCapacity
Capacity is how much sterilizing the world can do, where the plants are, who owns them, and what happens to the supply of devices when one closes.
6 · 2 · 2
Last achievedA breathing tube runs short2019
Watch forCobalt from Darlington
2
open
Advances to build with
2 research notes

Muse Ember
The jobs agents repeat
Meta's Muse Ember matched Claude Haiku 5.5 on an outside firm's tests at a third of the cost per task, though both list the same prices for text in and out.

Cursor Agent SDK
Findings that cite notes
Cursor released a kit for building AI agents, and its example, a code reviewer that cites the note behind each finding, is known only from Cursor's account.
Open in medical sterilization
3 of 13 open deliverables
Open · Ethylene oxide
Sterilization changes, approval by approval
A public, coded table of every FDA approval of a change to how a high-risk (Class III) device is sterilized. Each row gives the device, the date, the kind of change (a new sterilization site, a new contract sterilizer, an added chamber, a cycle change, less EtO, a step removed, a switch to another method), the method before and after where stated, and the sentence it was taken from. Yearly counts then show whether makers of high-risk devices have moved away from EtO since the FDA's 2019 challenges, or mostly moved from one EtO plant to another.
Built with what exists
the openFDA device PMA endpoint, which returns each supplement's type, reason and approval statement as JSON; a model that answers fixed-choice questions in bulk, such as Gemini Decision 1, to sort each statement into kinds of change (see the Note "A thousand questions, one pass"); DuckDB; two reviewers who know device regulation.
Open · Radiation
Dose-setting methods, rerun in public
An open program that carries out the published methods for setting a radiation sterilization dose (from a product's bioburden, from sterility tests at small incremental doses, and VDmax) and runs them on many thousands of made-up microbial populations whose resistance is known. It reports how often each method gives a dose too low to reach the target sterility assurance level and how often far too high, and lets anyone rerun the test for the lower doses of 15 to 20 kGy and for new resistance data. This is computer work only.
Built with what exists
Python with NumPy and SciPy; the published papers for each model's formulas; published survival curves of device bioburden under radiation for realistic populations; an AI coding agent to turn each paper's formulas into code with tests, each checked by a person against the paper.
Open · Steam and hospital sterile rooms
Sterile processing citations, sorted
A public table of every finding about sterile processing in US federal hospital inspection reports since 2011, covering cleaning, inspection, packaging, steam cycles, IUSS, indicators and records, storage, and staff training. Each row gives the step that failed, the quoted finding, the hospital, the date and a link to the report.
Built with what exists
HospitalInspections.org's text search; a model that answers fixed-choice questions in bulk, such as Gemini Decision 1, which scores up to 1,000 such questions in one pass (see the Note "A thousand questions, one pass"); the section headings of the 2025 multisociety guidance on sterilization (point-of-use treatment, IUSS, monitoring, tracking) as categories; two reviewers who know sterile processing.
Where it sits
Broader