Version control and locked reference sets for AI algorithms used as companion diagnostics
AI is starting to decide which patients get which cancer drug. Every change to the software should be tested against a fixed public set of cases before it is used on patients.
AI-based scoring of HER2, PD-L1 and other markers is entering clinical use. Software updates can shift positivity rates silently. Regulators (FDA's predetermined change control plans, EU AI Act) are building frameworks. A concrete requirement for oncology companion diagnostic algorithms: every version must report performance on a locked public reference set, changes must be logged with effect on positivity rates, and laboratories must record the algorithm version in each patient report.
- Biomarkers are not validated or standardised · Tests that decide who gets a drug are often not validated prospectively and are measured differently in every lab.
- AI that is built but not validated or deployed · Thousands of cancer AI models are published; a handful are in clinical use, and fewer have shown they help patients.
Pages like this
not linked directly; found by shared links- IdeaA standard evaluation pathway for AI-assisted pathology, from reader study to deployment
Shares PathAI, AI quantification of HER2-low and HER2-ultralow, Paige AI, Digital pathology & AI.
- IdeaAI second reads to stop borderline lesions being upgraded to cancer
Shares PathAI, Paige AI, Pathology & radiology foundation models, Digital pathology & AI.
- IdeaAI-first reading for high-volume common cancer diagnoses, pathologist for the exceptions
Shares PathAI, AI quantification of HER2-low and HER2-ultralow, Paige AI, Pathology & radiology foundation models.
- IdeaOne digital PD-L1 scale that maps across all the competing assays
Shares Public gold-standard datasets for validating every cancer biomarker test, PathAI, Paige AI, Digital pathology & AI.
- IdeaA registry of external validation datasets for cancer AI models, with mandatory reporting
Shares Public gold-standard datasets for validating every cancer biomarker test, Pathology & radiology foundation models, Digital pathology & AI, AI that is built but not validated or deployed.
- IdeaA pre-competitive consortium to train a shared multimodal cancer foundation model
Shares PathAI, Paige AI, Pathology & radiology foundation models, Digital pathology & AI.
- TechnologyPLUTO (PathAI)
- IdeaContinuous prospective validation for every oncology AI tool after deployment
Shares Pathology & radiology foundation models, Digital pathology & AI, AI that is built but not validated or deployed.