Digital health · Sepsis prediction
Sepsis Scout
A Y Combinator company that put a deep learning sepsis risk score on an off-the-shelf wearable and aimed it at cancer patients recovering at home. Acquired by Cytovale in 2024, it left a powerful founder story and a thin evidence file.
Eighteen bouts
Robert Quinn was 12 when doctors told him he would be dead before he turned 22 without a liver transplant. He got the transplant at 21. Then came the infections.
“Following the transplant, I suffered 18 separate bouts of sepsis,” he wrote on his company’s website. Sepsis is what happens when the body’s response to an infection turns on the body itself, and by the CDC figure the company cited it directly kills one time in six.
Quinn learned to read his own early signs, sore legs and a warm face, and would take himself to hospital before things turned. That is the origin story. Most people get no such warning, and most sepsis does not begin in a hospital bed with a monitor attached.
The part of sepsis nobody is watching
Hospitals have spent a decade wiring sepsis alerts into their records, which only cover patients already admitted. The company aimed at the other population: high-risk people at home between treatments, above all cancer patients whose chemotherapy has flattened their immune systems.
That framing is not marketing invention. A 2025 viewpoint in Lancet Primary Care, written by twenty sepsis researchers including Quinn, says wearables can flag people before clinical signs appear, then says in the next sentence that the trials to prove it have not been done.
Two engineers posting circuit boards to each other
Quinn studied engineering and biomedical science at the University of Queensland. The founder bio on the company’s Y Combinator page, written by its chief technology officer, describes electrical engineering and physiology, machine learning applied to electrophysiology in a neuroscience lab, then firmware and electronic design in implantable medical devices, where the two met. Aggregator records and the NIH grant name that co-founder as Wei-Jien Tan.
They started long distance, Quinn in Brisbane, Tan in Melbourne, mailing printed circuit boards to each other and meeting in Sydney every month or two. A $40,000 cheque from the Telstra-backed accelerator muru-D, as Quinn tells it, let them quit their jobs. When money ran short, Quinn drove for Uber at night and used every passenger to rehearse a sixty second pitch, which later won a St George Kick Start competition. The prize money and publicity opened a door to a hospital.
“The hospital then agreed to conduct the study with us,” Quinn told a University of Queensland podcast in 2020. The study got them into Y Combinator’s Winter 2018 batch, which got them to San Francisco. At demo day, TechCrunch described a chest-worn device, a paid pilot in Australia, 20 people wearing it, and a clinical trial planned for 2019.
Quinn was clear about why they moved. Regulators cost about the same either way, he argued, and the American market is far larger: “in Australia, that’s the Therapeutic Goods Administration in the United States, that’s the FDA or the Food and Drug Administration, but they’re roughly equal in terms of like money, time, energy committed to actually get it across the line.”
What the product actually was
By the time Patchd Medical became Sepsis Scout in 2023, the chest patch had become a watch. The archived technology page describes a device measuring heart rate, respiration rate, blood pressure, temperature and oxygen saturation, streaming through a phone to a cloud model that returns a risk score between zero and one. Cross the threshold and the care team gets an alert.
The model was a recurrent neural network that the company said looked for micro variations between vital signs, including diurnal rhythm, building what its FAQ called “a ‘fingerprint’ of sorts for sepsis”. The headline claim on the homepage was 39 hours of lead time. That number was never attached to a published study, a trial registration or a regulatory submission anywhere in the public record.
Federal money did back the algorithm. In 2022 the National Institute of General Medical Sciences awarded Sepsis Scout, Inc. a Phase I SBIR of $254,320, with Tan as principal investigator, to make the existing black box score explainable to clinicians. The grant ran to July 2024 and no Phase II follows it in NIH records. Quinn has also disclosed grant funding from BARDA.
The cautionary tale this category cannot avoid
In 2021, University of Michigan researchers externally validated the Epic Sepsis Model, a proprietary score running at hundreds of US hospitals, across 38,455 hospitalizations. It scored an area under the curve of 0.63, missed 67 percent of sepsis cases, and fired alerts on 18 percent of all hospitalizations.
That is the yardstick every sepsis algorithm now gets measured against, and it is why a missing external validation matters more here than it would in another category. Nothing in the record suggests this company’s model performed badly. Nothing in the record shows how it performed at all.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| FDA status | No 510(k), no PMA and no device listing under Sepsis Scout, Patchd or Patchd Medical in openFDA searches run Sep 23, 2026. A score that predicts a clinical event is a regulated function. | Not found |
| Registered trials | No study under either name on ClinicalTrials.gov. No Australian registration surfaced in searches. | Not found |
| Peer-reviewed validation | No PubMed paper reports the algorithm’s accuracy. A PubMed search for Patchd returns one unrelated 2009 record. | Not found |
| 39 hours of lead time | Stated on the homepage as a study result, with no study named, no cohort described and no publication. | Company-stated |
| NIH SBIR award | 1R43GM148091-01, NIGMS, $254,320, Aug 1, 2022 to Jul 31, 2024. PI Wei-Jien Tan. Purpose: add explainability to the existing prediction algorithm. No Phase II. | Public record |
| BARDA funding | Disclosed by Quinn in a 2025 Lancet Primary Care conflict of interest statement. Award numbers and amounts were not located. | Public record |
| Patents | US 2021/0052218 A1, inventor Robert Quinn, assignee Patchd, Inc., priority Aug 20, 2019. PitchBook also lists WO 2019/165004 A1, first filed Feb 21, 2018, and marks both inactive. No granted US patent found. | Public record |
| Australian pilot | TechCrunch reported in March 2018 a paid pilot in Australia and 20 people testing the device, with a trial planned for 2019. No results from either were published. | Independent |
| Clinical endorsement | The homepage carried a testimonial from Professor Simon Finfer, identified only by his Global Sepsis Alliance role. Finfer separately disclosed in JAMA (2024), Lancet Respiratory Medicine (2024) and Intensive Care Medicine (2025) that he owned stock options in Sepsis Scout. The site did not say so. | Not disclosed on the site |
| Acquisition | Confirmed in Quinn’s 2025 journal disclosure and by Y Combinator. PitchBook records the deal on Oct 23, 2024. | Public record |
Read plainly: a founder with an unusually direct reason to build this, real federal interest, a filed patent family, and no sign that anyone outside the company ever measured whether the algorithm worked. The category’s own literature says that measurement is the whole ball game.
What to watch
- Whether Cytovale, whose IntelliSep blood test was FDA cleared in December 2022, ever publishes or commercializes the home monitoring work it bought.
- Any peer-reviewed report of the 39 hour lead time claim, from any author.
- The patent family, listed as inactive, which would leave the method unprotected.
- Whether the NIH explainability work is published now that the grant has ended.
In their words
“The only reason that I am alive is that I can tell when I am getting sick.”
Robert Quinn, CEO and co-founder, company website · Company
“We quickly realized that if I could tell I was getting sick and save my own life, we could use deep learning and wearables to do the same thing for others.”
Robert Quinn, company website, archived 2023 · Company
“I had 18 episodes of sepsis following a liver transplant, actually.”
Robert Quinn, The Router podcast, UQ Computing Society, 2020 · Interview
“Patchd has built a chest-worn device that detects sepsis, the No. 1 cause of death in hospital patients.”
TechCrunch, Y Combinator W18 demo day, March 19, 2018 · Independent
“In studies, we’ve been able to detect sepsis, on average, 39 hrs earlier than is possible today.”
Company website homepage, archived 2023 · Company-stated
“and owning stock options in Sepsis Scout, unrelated to fluid therapy.”
Simon Finfer, disclosure, Lancet Respiratory Medicine, March 2024 · Public record
“This external validation cohort study suggests that the ESM has poor discrimination and calibration in predicting the onset of sepsis.”
Wong et al. on the Epic Sepsis Model, JAMA Internal Medicine, August 2021 · Peer reviewed
Related companies
Sources
- Public recordNIH RePORTER 1R43GM148091-01
- Public recordopenFDA searches, no matches
- Public recordClinicalTrials.gov searches, no studies
- Public recordUS 2021/0052218 A1, Systems and methods for sepsis detection and monitoring
- Public recordPreventing, identifying, and managing sepsis in the community
- Public recordExternal Validation of a Widely Implemented Proprietary Sepsis Prediction Model
- Public recordBLING III trial, author disclosures
- Public recordBEST-Living meta-analysis, author disclosures
- Public recordAdult sepsis in resource-limited settings, author disclosures
- Public recordCytovale IntelliSep 510(k) K220991
- IndependentHere are 64 startups that launched today at Y Combinator’s W18 Demo Day
- InterviewFrom UQ to Silicon Valley with Robert Quinn
- AggregatorSepsis Scout profile: funding, patents, acquisition date
- AggregatorSepsis Scout profile: investors and founders
- CompanySepsis Scout company page
- CompanySepsis Scout website: homepage, Our Story, Our Technology, FAQ, rebrand notice
- CompanyCytovale Completes $100 Million Series D Funding
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
