DoMore Diagnostics

A single deep learning model is the entire company. Histotype Px Colorectal reads standard H and E stained slides and returns a risk group; there is no assay, sequencing step, viewer, or laboratory of its own. The company's positioning makes the centrality explicit: the entire value proposition is extracting prognostic signal from slides already produced, in a field where the alternative biomarkers require gene sequencing that decades of investment have failed to deliver adequately for this indication.

Produces a low, intermediate, or high risk stratification that informs an oncologist's adjuvant chemotherapy decision, so the human retains the treatment call. The three tier output rather than a binary flag is a meaningful design choice, since it surfaces real uncertainty in the middle group rather than forcing every patient to one side of a threshold.

The consequential direction here is de-escalation, identifying patients who may safely avoid chemotherapy, which means a false low risk classification carries a different and more serious cost than a false high risk one, and the vendor materials do not quantify that asymmetry.

Training scale and provenance are disclosed, and the product is distinguished from its research predecessor. The company says the model was developed on 4,500 patient samples and 90 million image tiles, from slides prepared at several sites and scanned at two sites on two scanners, building on research led by the Institute for Cancer Genetics and Informatics at Oslo University Hospital; the early version was called DoMore-V1-CRC, and the CE-IVD certified product is an updated version. Naming that distinction, rather than letting the research paper stand for the shipping product, tells a buyer what the published evidence describes. Model architecture, version history and per subgroup performance are not published.

The approach and data named without the version and update discipline behind them.

An in-house model delivered through named host platforms, with the data path unstated. DoMore builds its own model, developed from research at the Institute for Cancer Genetics and Informatics at Oslo University Hospital, so no outside model provider sits between a slide and a risk group. The test is available on Sectra's Amplifier Marketplace, PathAI's AISight and Proscia's Concentriq Dx, so the laboratory's images sit in the host platform's environment. Nothing published says whether the model runs inside each host or whether images are sent to DoMore, which subprocessors apply, what DoMore retains, or whether customer slides contribute to model development; the privacy policy covers only the website. Whole slide image files embed a photo of the slide label by default, and with two parties in the path it matters which one strips it.

A structurally short chain at the model layer with the host platforms named, and the data path and terms unstated.

Peer reviewed development and validation, including a randomized trial cohort. An early version of the algorithm was published in The Lancet (a discovery and validation study), and Kleppe and colleagues in The Lancet Oncology showed that the marker improves prognostic accuracy when combined with TNM staging, validated in the QUASAR 2 trial with a hazard ratio of 9.47 between high and low risk groups; the company's site hosts both. Validation inside a randomized controlled trial cohort is a stronger standard than a retrospective institutional series, because trial cohorts have adjudicated outcomes and controlled treatment. Further clinical validations are running with Gustave Roussy, Erasmus MC, IFOM Milan (with ctDNA in the PEGASUS trial) and Ohio State, where a pilot is reported.

The research was led by the Institute for Cancer Genetics and Informatics at Oslo University Hospital, the company's origin, and the certified product is an updated version of the published algorithm.

Peer reviewed, multi cohort validation including a trial cohort, with method readable.

Nothing published describes how patient images are handled. The privacy policy covers personal data from website visits, email contact and job applications, with retention periods for those, and does not address slides or results from the product. The test runs through partner platforms, and nothing states whether DoMore receives customer slides for support or model improvement, what it retains, how images are de-identified, or how a laboratory is told if a model update would reclassify earlier cases. Whole slide files embed a photo of the slide label by default. No safety engineering for a wrong output is published.

Nothing published on how protected information moves through the product.

No position is published under HIPAA or its European equivalent for the product. The product carries a European IVD mark, and a US collaboration with Ohio State is a retrospective testing agreement, so a business associate position may not yet be needed. The non US route would instead need the European equivalent documented: a processor position and a data processing agreement for patient images analyzed through Sectra, AISight or Concentriq Dx, with the roles of DoMore and the host platform allocated. None is published. The only privacy policy covers personal data handled through the website, email and job applications, where DoMore is controller.

No status stated and no privacy document reaching the product.

Nothing independent is published for DoMore's own systems. The site publishes a privacy policy and a cookie policy only, with no SOC 2 report, ISO 27001 certificate, trust center or penetration testing statement. Because the test runs inside Sectra, PathAI and Proscia platforms, a laboratory's day-to-day security rests largely on the host, whose attestations cover the host, not DoMore. DoMore still touches data at three points: development and validation on real slides, support on failed cases, and model updates delivered into a regulated diagnostic pathway. The quality system behind the CE mark is not a security assessment.

Nothing independent published on security.

A European regulatory position stated, with no US clearance. The products page states that Histotype Px Colorectal is CE-IVDD marked and that the CE-IVD certified product is an updated version of the research algorithm (developed as DoMore-V1-CRC), indicated for stage II and III colorectal cancer to predict recurrence and survival from routine H&E slides. It is marked under the older IVD directive rather than the IVDR, which asks less clinical evidence of the manufacturer, and the product must move to the IVDR under the transition deadlines. FDA's 510(k) database lists no clearance for DoMore, and the US collaboration with Ohio State is described as a retrospective testing agreement.

A non US regime named, with the marketing matching it; no US clearance.

Validation in a trial cohort, and nothing published on performance across patient groups. Validation within the QUASAR 2 randomized trial gives controlled conditions and adjudicated outcomes, publication preceded regulatory approval, and the company says slides were prepared at several sites and scanned on two scanners at two sites. No results by patient group, site or scanner are published, and no governance framework or monitoring commitment is named. That gap matters for a de-escalation tool, since a model that underestimates risk in any group would steer those patients away from chemotherapy that would have helped them.

No evaluation by group published.

An oncologist weighs the risk group, and nothing stands behind it. Histotype Px places patients in low, intermediate or high risk groups to inform whether adjuvant chemotherapy follows surgery, and the treating team makes that decision. For a prognostic score the error that matters is calibration: predicted risk against observed recurrence in a stated cohort, with event counts. The company reports a hazard ratio of 9.47 between high and low risk groups in the QUASAR 2 trial cohort, which measures separation between groups, not how often a patient's group is wrong. Because the product's value is de-escalation, its consequential error is a patient placed at low risk who was not, whose chemotherapy may be withheld in a curable cancer, and that error cannot be checked against the slide. No warranty, indemnity or remediation commitment is published.

Clinical review of a decision support output with nothing attached to it.

Named host platforms, one of them a marketplace. The products page says Histotype Px Colorectal is available on Sectra's Amplifier Marketplace, PathAI's AISight and Proscia's Concentriq Dx, and the partners page describes the Sectra and PathAI collaborations; PathAI's own algorithm menu lists the product. That places the algorithm inside the image management systems laboratories already run rather than a separate environment. No integration documentation, LIS connection or production site by platform is published. For a pathology product the image management system and LIS, not the patient chart, are the correct surfaces.

Named systems with read access and results returned in the host viewer.

Delivered through host platforms, with location left to them. Histotype Px is available on Sectra's Amplifier Marketplace, PathAI's AISight and Proscia's Concentriq Dx, so a laboratory deploys it inside the image management system it already runs, and hosting, residency and retention follow that platform. Two architectures are possible behind the word integration: the model runs inside the host's environment and images never reach DoMore, or the host sends images to DoMore and receives a result. DoMore does not say which applies on each platform, and publishes no hosting location of its own.

A hosted route with location implied rather than committed.

No price is published; the value case is framed in the buyer's terms. The company states that most stage II and III colorectal patients are cured by surgery alone and that adjuvant chemotherapy benefits few of them, so the value of Histotype Px is avoided chemotherapy and toxicity, and it presents the test as giving risk information soon after surgery. The test is available on Sectra's Amplifier Marketplace, PathAI's AISight and Proscia's Concentriq Dx, and the site has an order page, but no fee, per test rate or channel terms are published.

The sales posture is discoverable and no price is published.

A single indication, precisely stated and validated in a trial cohort. Histotype Px Colorectal predicts recurrence and survival in stage II and III colorectal cancer after surgical resection from routine H&E slides, to inform the adjuvant chemotherapy decision. The company says it was developed on 4,500 patient samples and 90 million image tiles, with slides prepared at several sites and scanned at two sites on two different scanners, and it was validated in the QUASAR 2 trial cohort (The Lancet Oncology). Clinical validation studies are running with Gustave Roussy, Erasmus MC (up to 2,200 rectal cancer patients), IFOM Milan and Ohio State. It does not extend to other tumor types or metastatic disease, and the company does not name the scanners.

Neither the scanner nor the population detail is given for laboratories outside the study settings.