Cytovale IntelliSep

Machine learning is essential but it is not the moat, and the distinction matters for how a buyer should evaluate this. The company states that the system uses high throughput image analysis, computer vision and machine learning to examine biologically relevant features in cell images and translate them into a score, and independent trade coverage describes the test the same way. So the model is what converts tens of thousands of cell images into a clinical probability, and without it there is no result. But the differentiated asset is the physical measurement technique, microfluidic deformability cytometry, together with the biological insight that leukocyte biomechanics shift with immune activation. The model is the readout layer on top of a hardware and biology platform. Same structural position as Eko Health, which is also graded B for being coupled to its own hardware.

THE INTERMEDIATE BAND IS A PUBLISHED ABSTENTION MECHANISM AND IT DESERVES EXPLICIT CREDIT, because this index asks nearly every vendor what their system does when it does not know and almost never gets an answer. Band 2, covering scores from 5.0 to 6.2, exists precisely to say the result is indeterminate rather than forcing a binary call, and the score ranges for all three bands are published so a clinician can see exactly where the boundaries sit. The adjunctive constraint is stated plainly and repeatedly: results are intended for use in conjunction with clinical assessment and other laboratory findings, and all results should be interpreted in the context of other clinical observations. Nothing is auto actioned. Held at B rather than A because no performance characteristics per band are published, so a clinician knows where the thresholds are but not what a Band 3 result actually implies in terms of probability, and because the low probability band carries the same quiet de escalation risk this index records across the deterioration category.

The technology stack is described in unusual detail and the biology is published, which together carry this above the market norm. Named components: precision microfluidics, deformability cytometry, ultra high resolution imaging at very high frame rates, high throughput image analysis, computer vision and machine learning. The biophysical mechanism is not asserted but cited, including Di Carlo's work on mechanical biomarkers of cell state and a 2023 paper on biophysical changes of leukocyte activation and NETosis in the cellular host response to sepsis. The FDA summary describes the sample workflow concretely, down to the 100 microlitre volume and the preparation steps. WHAT IS ABSENT IS THE MODEL ITSELF: no architecture, no feature list, no training population, no calibration, and no account of how the model was validated separately from the assay as a whole. For a product where the model performs the actual classification, that is the gap between this grade and an A.

Clearance rests on the CV-SQuISH-ED multicentre clinical validation study, with a named national principal investigator, Hollis O'Neal MD of Our Lady of the Lake Regional Medical Center and LSU Health Sciences Center, which is a stronger evidentiary basis than most 510(k) submissions in this index carry. Independent evaluation exists: researchers at the University of Kansas Medical Center assessed the test against monocyte distribution width, von Willebrand factor and ADAMTS13 in 44 emergency department patients, published in the American Journal of Clinical Pathology. Peer reviewed mechanism papers support the underlying biology. Held at B rather than A on three counts: the pivotal validation study was described as being prepared for publication and its published status could not be confirmed in this pass, the independent evaluation was small at 44 patients and appeared as a supplement abstract, and no sensitivity, specificity or predictive value figures per band were retrieved, so the operating characteristics behind the three band structure are not publicly available even though the bands themselves are.

No privacy policy, data processing statement, retention period or training use disclosure was located. Not Rated means no public evidence was found rather than evidence of weakness. The exposure profile here is narrower than for software vendors in this space, since the instrument runs inside the hospital laboratory on a physical blood sample and the primary output is a numeric result entering the laboratory information system. The open question is whether cell image data or results are transmitted to the vendor for quality monitoring or model improvement, which nothing published addresses.

No public statement on business associate agreements, execution terms or subprocessor disclosure was located. Not Rated means no public evidence was found. As an in vitro diagnostic run on hospital owned instrumentation by hospital laboratory staff, the contracting posture differs from a software vendor's and a buyer should establish whether any vendor data connection exists at all before assuming an agreement is required.

No SOC 2 of any type, HITRUST, ISO 27001, trust centre or report request path was located. Not Rated means no public evidence was found. As recorded throughout this index, FDA 510(k) clearance is a safety and effectiveness determination and is not an information security attestation. For connected laboratory instrumentation the relevant questions are medical device cybersecurity and network posture rather than cloud attestations, and neither is addressed publicly.

First in class clearance with a precisely bounded indication, which is the combination this axis rewards. IntelliSep received 510(k) clearance under K220991 in December 2022 and is described as the first and only FDA cleared host response diagnostic indicated for use in the emergency department. THE INDICATION IS UNUSUALLY SPECIFIC AND THAT SPECIFICITY IS A VIRTUE: adult patients with signs and symptoms of infection presenting to the emergency department, to aid early detection of sepsis WITH ORGAN DYSFUNCTION MANIFESTING WITHIN THE FIRST THREE DAYS AFTER TESTING. Naming the outcome, the population, the setting and the prediction window in the cleared indication itself tells a buyer exactly what was demonstrated, which is far more than the open ended deterioration claims common in this space. The three interpretation bands and their score ranges are published alongside it.

No subgroup performance, calibration or fairness analysis is published. The exposure specific to this technology is unexamined and is worth naming precisely because it is biological rather than demographic: the test infers immune activation from leukocyte biomechanics, and leukocyte populations and behaviour are altered by haematological disease, active chemotherapy, immunosuppression, chronic inflammatory conditions, corticosteroid use and age. Any of these could plausibly shift the index independently of sepsis, and nothing published reports performance in those groups or states them as limitations. A conflict disclosure is recorded neutrally and to the company's credit for being visible: the originating researcher and the Regents of the University of California hold financial interests in Cytovale, disclosed in the peer reviewed literature.

No EHR integration, laboratory information system interface, HL7 or FHIR capability was described anywhere retrieved. As a laboratory instrument the expected path is a result flowing through the LIS into the record, and the company emphasises easy integration into existing emergency department processes, but the mechanism is not documented. That matters more than it would for a batch laboratory test, because the entire value proposition depends on an eight minute result reaching a clinician inside a time critical triage decision, and how the result surfaces is therefore part of whether the product works. Establish the interface path and where the result appears in the emergency department workflow.

A self contained instrument model with genuine operational advantages and one clear cost. The test runs on the Cytovale System inside the hospital, requires only 100 microlitres of whole blood in a standard EDTA tube, involves automated sample preparation rather than manual processing, and returns a result in roughly eight to ten minutes, which is fast enough to sit inside an emergency department triage decision rather than after it. Named deployment context includes Our Lady of the Lake Regional Medical Center, and a 100 million dollar Series D was raised explicitly to fund commercial expansion into more emergency departments. Held at B because it requires dedicated capital equipment and consumable cartridges in the laboratory rather than being a software deployment, no installed base or site count was retrieved, and no implementation timeline or laboratory prerequisite is published.

No pricing published: no instrument cost, no per test or cartridge price, no band and no implementation fee. That gap is more consequential than for software because the model carries both capital equipment and per test consumables, so a buyer faces two distinct cost lines and neither is disclosed. ONE PARTIAL DISCLOSURE IS CREDITED WITHOUT LIFTING THE GRADE: the company has applied for a CPT Proprietary Laboratory Analyses code, which is the reimbursement half of the equation and the same category of disclosure that earned Eko Health and HeartSciences credit elsewhere in this index. It falls short of their position because no code, status or payment rate was retrieved, so a buyer cannot yet model the offset.

Deliberately narrow and explicitly bounded, which this index treats as a strength rather than a limitation when the boundaries are published. Coverage is adult patients with signs and symptoms of infection presenting to the emergency department, with a stated prediction window of organ dysfunction within three days of testing. The setting choice is strategically coherent, since the company cites more than 80 percent of sepsis cases as presenting through the emergency department, and the test is positioned for the triage moment specifically rather than for ongoing monitoring. Held at B because there is no paediatric indication, no inpatient or intensive care indication, and no application outside sepsis, and because the company describes sepsis as the first application of the underlying platform without documenting any second one.