It stopped asking whether you have an infection - and started asking what your immune system already knows.
In an emergency department somewhere right now, a clinician is staring at a patient who looks unwell and faces the oldest question in acute medicine: is this a bacterial infection that needs antibiotics this hour, a virus that does not, and how close is this person to falling off a cliff? For decades the honest answer was a shrug dressed up as a blood culture - results in a day or two, long after the decision had to be made. Inflammatix built the test that answers in about thirty minutes, and it does so by ignoring the germ entirely.
The company's FDA-cleared TriVerity test reads 29 of the patient's own genes - the mRNA signature of an immune system mid-fight - and runs them through machine-learning algorithms trained on over a decade of data. Out come three scores: how likely the infection is bacterial, how likely it is viral, and how severe the illness is going to get. It runs on a benchtop instrument called Myrna, named, with the kind of pun scientists rarely resist, after the mRNA chemistry inside.
Sepsis is a runaway immune reaction to infection, and it is one of medicine's cruelest stopwatches: survival drops with every hour of delay. The tools clinicians inherited were built to find the pathogen - culture it, amplify its DNA, match it to a name. Useful work, but slow, and slowness is exactly the enemy. So clinicians do what reasonable people do under uncertainty: they over-treat. Antibiotics get handed out to people who do not need them, and the people who genuinely need them sometimes wait too long.
The result is a system that is simultaneously too aggressive and too cautious. It feeds antibiotic resistance on one end and misses deteriorating patients on the other. Inflammatix's founders looked at this and noticed something the rest of the field had largely walked past: the patient's own body had already diagnosed the problem. The immune system reacts to a bacterial threat differently than a viral one, and it does so within hours. Nobody was reading that report.
In 2016, Tim Sweeney - then a Stanford surgery resident with a PhD in computational immunology - and Purvesh Khatri, the Stanford researcher behind the underlying science, spun the idea out into a company alongside co-founder Jonathan Romanowsky. Their wager was unfashionable. The diagnostics industry had spent years and fortunes perfecting pathogen detection; proposing to read the host instead drew, in Khatri's own telling, "deep skepticism."
Khosla Ventures took the early bet anyway, and the science had to do the rest. The bet was not really about one test. It was about a platform: if the immune response is a readable, machine-interpretable signal, then a benchtop instrument and a cartridge could turn that signal into a clinical decision at the bedside. That is a different business than chasing one bug at a time - and a harder one to fund, which is part of why so much of the company's capital arrived as patient, science-stage money.
Co-Founder & CEO. Former Stanford surgery resident, MD/PhD in computational immunology, and the public face of the host-response argument.
Co-Founder & Chief Scientist. The Stanford researcher whose multi-cohort gene-signature work is the engine under the hood.
Co-Founder and board member, part of the original 2016 team that turned a research idea into a company.
TriVerity is the test; Myrna is the box that runs it. A blood sample goes into a cartridge, the cartridge goes into the Myrna instrument, and isothermal RT-LAMP chemistry amplifies the 29 target mRNAs without the bulky thermal cycling older methods require. Roughly half an hour later, the clinician has three numbers instead of a maybe.
Likelihood the infection is bacterial - the score that decides whether antibiotics are the right call now.
Likelihood the infection is viral - the score that helps clinicians safely not reach for antibiotics.
All-cause risk of needing ventilation, vasopressors, or renal replacement within seven days. The deterioration early-warning.
It is, deliberately, a sample-to-answer point-of-care system rather than a send-out lab assay. The whole design philosophy is that the answer should arrive while the decision is still open - a modest-sounding goal that the existing toolkit had quietly failed to meet for years.
A bold idea is just a press release until the data lands. Inflammatix's pivotal SEPSIS-SHIELD study enrolled 1,222 patients across 22 sites, and the headline finding is the one that matters at 3 a.m.: high diagnostic and prognostic accuracy that held regardless of a patient's immune status or race. Stanford Medicine reported a roughly 0.5% false-negative rate, against something closer to 2% for standard practice. Small-sounding percentages; large-sounding consequences.
| Round | Amount | Key investors |
|---|---|---|
| Series C (Jan 2020) | $32M | Khosla Ventures, Northpond, Think.Health, Grey Sky |
| Series D (Mar 2021) | $102M | D1 Capital Partners (lead), Northpond, Khosla, OSF HealthCare Ventures |
| BARDA grant (Nov 2021) | $12.1M | BARDA (non-dilutive, ViraBac EZ) |
| Series E (Sept 2024) | $57M | Khosla & Think.Health (lead), Northpond, D1, Iberis, Vesalius, OSF, RAW |
Partnerships fill in the rest of the picture: BARDA for development funding, OSF HealthCare as both investor and health-system partner, Stanford as the source of the science, and the U.S. military's JPEO-CBRND as an early customer - an initial $1.2M order with phase options reportedly up to $20.8M. When the Department of Defense wants your benchtop instrument in the field, the host-response thesis has graduated from interesting to operational.
Inflammatix describes its mission as helping bring about a new era in emergency and critical care - a future where urgent care is as responsive and effective as it can be. Stripped of the ceremony, it means this: clinicians should get to act on the body's own signal instead of waiting on a culture plate. The company's one-line framing - "interpreting the immune response to better diagnose disease" - is less a slogan than a job description.
The business model follows the mission's logic. Place Myrna instruments where acute decisions get made, sell the per-test TriVerity cartridges, and lean on non-dilutive government funding to underwrite a science-heavy roadmap. It is the razor-and-blade model with a regulatory moat and a decade of immunology behind the blade. The competition - Cytovale, Prenosis, MeMed, Immunexpress, and the entrenched habit of lactate, procalcitonin, and culture - is real, which is precisely why the depth of validation matters.
Antibiotic resistance is the slow-motion crisis that every public-health body keeps circling. Tests that tell clinicians when not to prescribe are among the few levers that work on it. A test that also flags who is about to crash adds the other half of the equation - fewer missed deteriorations, fewer unnecessary prescriptions. If host-response diagnostics become standard in the emergency department, the win is not one test sale; it is a shift in how acute medicine reasons under uncertainty.
Return to that emergency department from the opening. The clinician still has a patient who looks unwell and a decision that cannot wait. The difference is what is now on the screen: three numbers, arrived in about thirty minutes, drawn from the patient's own immune system instead of guessed at from a fever and a gut feeling. Inflammatix did not invent the question. It just made the body answer it out loud - and made the clock, for once, the doctor's ally.