Validation and regulatory status
How each product is validated against pathologist reads, how model versions are controlled, and where each product stands with regulators.
How products are validated
Reference standard
Pathologist annotation, compared reader by reader, or a molecular reference where one exists.
Held-out cohort
Cases from sites and scanners not seen in training; at least 100 per product.
Metrics that match the use
Agreement at the clinical threshold, not only overall accuracy.
Locked and versioned
Each release is frozen, numbered and documented before it reaches a laboratory.
Status by product
Status is stated per product and per jurisdiction, in the wording of the applicable mark.
Product | Regulatory status | Validation summary |
|---|---|---|
| Octo Cellularity | Research use only | 83.1% agreement with consensus · 414 cases |
| Octo Immune | Research use only | 942 cases |
| Octo Mitosis | Research use only | 86% detection accuracy · 105 images |
| Octo E-N | Research use only | Validation in progress |
| Octo Duct | Research use only | 128 resections · gland architecture prognostic in TCGA-PAAD |
| Nemo ER-PR | Research use only | Detection F1 0.84 · 210,419 nuclei |
| Nemo HER2 | Research use only | F1 0.984, pilot |
| Nemo PD-L1 | Research use only | Detection F1 0.82 · TPS r = 0.93 · 32 regions |
| Nemo Ki67 | Research use only | F1 0.934, pilot |
Quality and security
Quality management
ISO 13485 quality system in preparation.
Information security
NHS DSPT submitted; ISO 27001 in preparation; on-premises by default.
Post-market surveillance
Performance monitored per site; drift reported.
Need the documentation?
Validation summaries and technical files are shared with laboratories and partners under NDA.