# A dedicated fund for randomised trials of cancer AI with patient outcomes

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

## TL;DR

Thousands of cancer AI tools have been tested on old data; almost none in a proper trial. Fund the trials, with endpoints that matter to patients.

## Summary

Systematic reviews find that fewer than a few percent of published oncology AI models have prospective evaluation and vanishingly few have randomised trials with clinical endpoints. Exceptions such as the MASAI mammography screening trial in Sweden show the design is feasible and informative. The proposal is a public and philanthropic fund that pays for pragmatic randomised trials of AI tools in pathology, radiology, screening and decision support, requiring pre-registration, clinical endpoints (cancer detection rate, interval cancers, time to treatment, survival) and open reporting.

## Fields

- Kind: Idea
- Last checked: 2026-09-08
- Hypothesis: Randomised evaluation will show that a minority of AI tools with strong retrospective performance improve patient-relevant outcomes, and the resulting evidence will drive adoption of those that do and retirement of those that do not.
- Rationale: Retrospective accuracy has repeatedly failed to translate into benefit in other digital health interventions; only randomisation resolves the question, and the MASAI trial shows it can be done at scale within a screening programme.
- Proposed test: Fund ten pragmatic randomised trials of deployed or near-deployed cancer AI tools over five years; report how many show benefit, harm or no effect.
- Maturity: early-clinical
- Actor: philanthropy

## Sources

- MASAI trial (Lancet Oncology 2023): https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(23)00298-X/fulltext

## Connected records

- fronts: [AI & Computation](https://onco.cc/fronts/ai-computation/)
- technologies: [AI in radiology](https://onco.cc/technologies/radiology-ai-screening/), [Digital pathology & AI](https://onco.cc/technologies/digital-pathology-ai/), [Mammography & tomosynthesis](https://onco.cc/technologies/mammography/)
- bottlenecks: [AI that is built but not validated or deployed](https://onco.cc/bottlenecks/b-ai-validation/)
- ideas: [Pay for cancer AI only when it has outcome evidence, then pay properly](https://onco.cc/ideas/idea-data-ai-reimbursement-tied-to-outcomes/)

---
JSON: https://onco.cc/api/v1/entities/idea-data-prospective-ai-trials-fund.json