Viz.ai

Detection models sit under everything, with more than 50 FDA cleared algorithms analyzing CT, ECG, and echocardiography. But the product's distinguishing move is what happens after detection: the system automatically alerts and connects the treating specialist rather than only marking a study. Both halves are essential, and neither works without the model identifying the finding first.

The regulatory category defines the boundary precisely and conservatively. The original LVO product received FDA De Novo authorization as a computer aided triage and notification platform, meaning it notifies a specialist earlier rather than rendering a diagnosis, and the FDA's own framing at authorization was that earlier notification could decrease time to treatment. In practice the alert reaches a physician's phone with the images attached, so the human sees the underlying evidence and makes the call. For time critical stroke care, compressing notification while preserving physician judgment is the correct design.

The original authorization came with disclosed performance: an AUC of 0.91 in a 300 patient study, 90 percent sensitivity and specificity, and a median scan to notification time under six minutes. Subsequent clearances specify indications concretely, such as quantifying intracranial hyperdensities, lateral ventricles, and midline shift on non contrast CT. Architecture is described as deep learning without further detail, and per algorithm performance across the 50 plus cleared products is not published in aggregate.

Among the strongest outcome evidence in imaging AI, and unusually it measures care delivery rather than detection accuracy. Peer reviewed studies in Interventional Neuroradiology and the Journal of NeuroInterventional Surgery document reduced transfer times and length of stay and improved stroke workflow in hub and spoke networks, with the company citing decreases in door to transfer, door to puncture, and door to recanalization alongside improved Modified Rankin Scores, the standard disability outcome measure. Independent analysis has reported higher LVO detection rates with AI than without. Deployment spans over 1,700 hospitals. Moving the outcome metric from image reading to patient disability is what separates this from most of the category.

The company states HIPAA compliant image review within its coordination platform, which matters because the workflow deliberately pushes patient imaging to physicians' mobile devices, a broader exposure surface than PACS resident AI. No detailed data governance, retention, or training data disclosure was located beyond that compliance statement.

HIPAA compliant image review and care plan discussion are stated for the mobile coordination platform. No explicit business associate agreement commitment was located, which is what would move this higher given the volume of identifiable imaging moving through the system.

No SOC 2, HITRUST, or ISO 27001 attestation was located, and no trust center was found. For a platform distributing patient imaging to mobile devices across 1,700 hospitals, published security attestation would be a reasonable expectation.

Historically significant and currently the broadest in imaging AI. The company received the first ever FDA De Novo authorization for computer aided triage and notification software in February 2018 under DEN170073, creating the regulatory category that many competitors now clear into, followed by CE marking in Europe. It now reports more than 50 FDA 510(k) clearances spanning stroke, intracranial hemorrhage quantification, pulmonary embolism, and aortic disease. Buyers should confirm which specific modules and indications their contract covers, since clearances are per algorithm.

No governance framework or bias evaluation was located. The equity question here is structural rather than demographic: a platform that accelerates transfer from spoke to hub hospitals allocates access to thrombectomy capacity, and whether that acceleration benefits all catchment populations equally is not addressed in published materials.

Integration is imaging and communication rather than chart based, and the communication layer is the differentiator. The platform connects directly to hospital CT scanners via DICOM and pushes findings through a proprietary mobile application that serves as the coordination surface for neurologists, interventional radiologists, and stroke nurses across primary and comprehensive stroke centers. Spanning multi hospital networks is the harder version of this problem. No EHR integration was located.

No hosting, tenancy, or data residency terms were located. The architecture necessarily moves imaging out of the hospital to reach mobile devices, which makes the absent residency disclosure more consequential here than for PACS resident products.

No published pricing, and third party analysis describes enterprise only contract based licensing that varies by hospital size, module set, and term, typically structured either per indication or as a platform wide subscription with multi year commitments. One genuinely useful commercial fact does exist publicly: reimbursement through the Medicare New Technology Add on Payment pathway has applied in stroke care, which materially changes the return calculation and is worth a buyer investigating for their own case mix.

Broad and deep simultaneously. Clinical coverage spans neurovascular disease including LVO stroke and intracranial hemorrhage, pulmonary embolism, aortic disease, and cardiac indications, across CT, ECG, and echocardiography. Setting coverage is the more distinctive dimension: the platform is built for hub and spoke networks connecting primary stroke centers to comprehensive centers, which is a system level rather than department level deployment, across more than 1,700 hospitals globally.