# Version control and locked reference sets for AI algorithms used as companion diagnostics

Source: https://onco.cc/ideas/idea-tr2-ai-cdx-change-control/  
OnCo record `idea-tr2-ai-cdx-change-control` (Idea). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

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.

## Summary

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.

## Fields

- Kind: Idea
- Last checked: 2026-09-08
- Hypothesis: Version reporting will reveal at least one clinically meaningful drift (positivity change above five percentage points) in a deployed algorithm within two years, which would otherwise have gone undetected.
- Rationale: Software versioning is routine in engineering and absent in diagnostic pathology reporting; drift has been documented in deployed medical AI.
- Proposed test: Implement version logging and reference-set testing for two deployed pathology algorithms across ten laboratories; monitor positivity rates by version.
- Maturity: early-clinical
- Actor: regulator

## Sources

- Bottleneck evidence (Biomarkers are not validated or standardised): Fernandez et al., Examination of low ERBB2 protein expression in breast cancer tissue (JAMA Oncology 2022): https://doi.org/10.1001/jamaoncol.2021.7239

## Connected records

- companies: [Paige AI](https://onco.cc/companies/paige/), [PathAI](https://onco.cc/companies/pathai/)
- ideas: [AI quantification of HER2-low and HER2-ultralow](https://onco.cc/ideas/idea-ai-her2-low-scoring/), [Monitor biomarker positivity rates across labs in real time to catch assay drift](https://onco.cc/ideas/idea-tr2-positivity-rate-surveillance/), [Public gold-standard datasets for validating every cancer biomarker test](https://onco.cc/ideas/idea-tr2-open-cdx-validation-sets/)
- technologies: [Digital pathology & AI](https://onco.cc/technologies/digital-pathology-ai/), [Pathology & radiology foundation models](https://onco.cc/technologies/pathology-foundation-model/)
- bottlenecks: [AI that is built but not validated or deployed](https://onco.cc/bottlenecks/b-ai-validation/), [Biomarkers are not validated or standardised](https://onco.cc/bottlenecks/b-biomarker-validation/)
- key papers: [Examination of Low ERBB2 Protein Expression in Breast Cancer Tissue](https://onco.cc/key-papers/paper-fernandez-jama-oncol/)

---
JSON: https://onco.cc/api/v1/entities/idea-tr2-ai-cdx-change-control.json