Aidoc

Aidoc is software a hospital installs between its imaging systems and its clinicians. Studies arriving from CT, X-ray and echo flow through the platform, algorithms look for specific findings, and positive cases are marked in the radiologist's worklist and pushed to the relevant care team. The vendor describes multiple algorithms running across a single scan with the results presented in the context of the patient record, which is why one study can be checked for intracranial hemorrhage, pulmonary embolism and a vertebral fracture at once.

Radiologists see the output through what Aidoc calls the Widget, a panel that consolidates results from its own and partner algorithms, runs on any workstation and marks triage results directly in the worklist. Downstream clinicians get the same case on mobile: care coordination alerts the right users and lets them chat and share images about a patient who has not yet been read.

The pathology list is specific rather than general-purpose. Published groupings cover neurovascular (aneurysm, intracranial hemorrhage, midline shift, vessel occlusions, C-spine and vertebral fractures), aortic, cardiac, venous thromboembolism, chest, and a long abdominal set including appendicitis, bowel obstruction and obstructive kidney stone.

aiOS is sold as the layer that makes clinical AI deployable rather than as a single detector. Aidoc says 69% of customers already run partner or homegrown algorithms on it, and the partner page names the third-party developers whose applications are delivered through it: Gleamer, Icometrix, Imbio, ScreenPoint, Riverain, Us2.ai, Circle CVI and Cercare Medical.

Integration is described as a vendor-neutral back-end that connects over DICOM, HL7 and FHIR to PACS, EHR, VNA, scheduling, worklist and mobile communication systems, with sites live in two to three weeks and most setup done remotely. On the same page Aidoc claims to be the only AI vendor available in Epic's App Orchard.

Governance sits in the same product: aiOS reports validation, drift detection, override tracking and analytics so an AI committee can see what the models did and when clinicians disagreed with them.

Rather than calling an external model provider, Aidoc trains its own. CARE, the Clinical AI Reasoning Engine, is described as a clinical-grade foundation model trained on real-world multimodal data that already powers cleared applications, with report generation from imaging pixels named as a direction of travel.

That shapes what model questions mean here. A buyer does not pick between vendors of general language models; the choice available is which cleared imaging algorithms, Aidoc's own or a partner's, run on the platform. Nothing on the site addresses open-weight models or supplying your own provider key, and no page claims either is possible.

There is no pricing page, no plan table and no figure anywhere on the site. The purchase question is answered plainly: reach out to the team, or buy through Change Healthcare, Merative or Radiology Partners. For a hospital procurement team that is a normal shape; for anyone comparing cost it means the entry price, the unit of pricing and whether partner algorithms carry their own fee are all unpublished.

There is also no free tier and no trial. The only self-service artefacts are gated PDFs and webinars.

The clinical and quality side is documented well. Aidoc lists MDSAP certification, ISO 13485:2016, EU 2017/745 and compliance with FDA 21 CFR Part 820, alongside FDA, TGA and CE clearances, and points readers at its 510(k) notifications. Independent clinical literature exists: the site cites peer-reviewed turnaround-time and accuracy studies rather than only its own case notes.

Information security is asserted at a higher level. The security page names a CISO-led team, the NIST Cybersecurity Framework, hosting on AWS and Azure, and sends buyers to a Trust Center for documents; the integration page claims SOC2, GDPR, HIPAA compliant. What is not published anywhere public: an uptime commitment, identity-provider sign-in support, retention controls for scan data, or a region choice for storage.

One gap is worth naming precisely. The privacy policy states that Customer Information uploaded to the platform "is not regulated by this privacy policy", so the clinical data handling a hospital most wants in writing is left to the contract. The FAQ's claim that "no patient health information leaves your facility at any time" is the closest public statement, and it sits in an FAQ rather than a data-processing document.

This suits a health system that already has PACS and an EHR, an AI governance committee, and more imaging volume than its radiologists can prioritize by hand. Aidoc's references are at that scale: 150-plus health systems, deployments across 22 sites for one nonprofit network, 17 cleared algorithms live at another.

It is the wrong product for an individual clinician, a small outpatient practice buying one tool, or a developer who wants programmatic access to imaging AI. There is no sign-up, no published price, no API reference, and the value depends on integration work done with hospital IT. Teams whose procurement needs a written SLA or an SSO answer before signing will have to extract both from the Trust Center or the contract, because neither appears on the public site.