AI in the oncology clinic: from narrow cleared tools to multimodal decision support
How AI is moving from single-task readers of scans and slides towards systems that weigh everything about a patient, and what regulators and evidence still require.
Hundreds of narrow AI devices are cleared, mostly in radiology triage and screening. The first predictive pathology tools (ArteraAI) prove that AI can change treatment decisions under regulation. Foundation models promise breadth, but the evidence base for deployment, the regulatory framework for updating models, and reimbursement are all unsettled.
- 2017-2021historic
Narrow detection tools cleared
FDA clears the first AI detection aids: mammography CAD successors, Paige Prostate (2021), radiology triage for haemorrhage and embolism. Evidence is mostly reader studies.
- 2022-2024current
Screening at scale and risk models
MASAI (Sweden) shows AI-supported mammography reading finds more cancers with less workload; Sybil and Mirai predict future cancer from today's scan; whole-slide imaging becomes routine, enabling slide-level AI.
- 2025-2026current
First predictive AI and foundation models in products
ArteraAI Prostate (de novo 2025) predicts treatment benefit; ArteraAI Breast cleared 2026. Pathology foundation models (Virchow2, Prov-GigaPath, TITAN, Atlas) move into commercial biomarker products; multimodal models (MUSK) predict immunotherapy response retrospectively.
- 2026-2028emerging
Prospective evidence and regulatory frameworks
Randomised or pragmatic trials of AI-guided decisions (screening intervals, treatment selection); FDA predetermined change control plans for model updates; EU AI Act high-risk obligations; payment codes for AI-derived biomarkers. LLM assistants (Med-Gemini class) enter tumour boards for documentation and trial matching under human review.
- 2029+speculative
Speculative: multimodal decision support as standard of care
A single model reads slides, scans, genomics and records to recommend and monitor therapy, audited against outcomes and updated continuously. Depends on data-sharing, liability and validation questions that are open today.
Probability ranges are named estimates that the claim is borne out on roughly a five-year horizon. They are meant to be argued with: propose a revision with your name and reasoning via a pull request to src/data/confidence.ts.
Story
topNarrow detection tools cleared
FDA clears the first AI detection aids: mammography CAD successors, Paige Prostate (2021), radiology triage for haemorrhage and embolism. Evidence is mostly reader studies.
Scanning microscope slides and letting software measure things a pathologist cannot see, including predictions of who will benefit from a treatment.
Software that reads scans alongside radiologists, catching cancers earlier and predicting who is at risk.
MSK spin-out with the first FDA-cleared AI pathology product and the Virchow foundation model.
Screening at scale and risk models
MASAI (Sweden) shows AI-supported mammography reading finds more cancers with less workload; Sybil and Mirai predict future cancer from today's scan; whole-slide imaging becomes routine, enabling slide-level AI.
Low-dose breast X-ray used for screening. Newer 3D versions find more cancers with fewer false alarms.
Predicts a person's six-year lung cancer risk from one low-dose CT, even when no nodule is visible.
Reads a mammogram to estimate five-year breast cancer risk, consistently across races and devices.
Scanning microscope slides and letting software measure things a pathologist cannot see, including predictions of who will benefit from a treatment.
First predictive AI and foundation models in products
ArteraAI Prostate (de novo 2025) predicts treatment benefit; ArteraAI Breast cleared 2026. Pathology foundation models (Virchow2, Prov-GigaPath, TITAN, Atlas) move into commercial biomarker products; multimodal models (MUSK) predict immunotherapy response retrospectively.
The first AI tool cleared by the FDA to predict both prognosis and treatment benefit from a routine biopsy slide, in prostate cancer.
An FDA-cleared AI test (May 2026) that reads breast cancer slides to estimate recurrence risk in early hormone-positive disease.
A pathology foundation model trained on millions of slides that can detect cancer and predict biomarkers from an ordinary H&E slide.
An open pathology model trained on 1.3 billion image tiles from a US health system, modelling whole slides at gigapixel scale.
TITAN is a model that summarises a whole slide, not just tiles, and can write a draft pathology report.
Atlas is a pathology foundation model trained on 1.2 million slides from two of the world's largest hospitals.
A model that reads slides and clinical text together to predict who will respond to immunotherapy.
Prospective evidence and regulatory frameworks
Randomised or pragmatic trials of AI-guided decisions (screening intervals, treatment selection); FDA predetermined change control plans for model updates; EU AI Act high-risk obligations; payment codes for AI-derived biomarkers. LLM assistants (Med-Gemini class) enter tumour boards for documentation and trial matching under human review.
Google's medical versions of its Gemini models, able to reason over text, images, and long records.
Software, increasingly LLM-based, that reads a patient's record and finds trials or guideline options they qualify for.
Models trained on Tempus's paired genomic, pathology, imaging and outcome data to predict response and prognosis.
A model trained on millions of hospital records that forecasts a patient's next diagnoses.
Speculative: multimodal decision support as standard of care
A single model reads slides, scans, genomics and records to recommend and monitor therapy, audited against outcomes and updated continuously. Depends on data-sharing, liability and validation questions that are open today.
Train one AI on scans, slides, genomics, and outcomes from many patients so it can predict, for a new patient, which treatment will work.
Real-world evidence at scale: what happened to patients with a given genomic profile on a given treatment.
The Imaging Data Commons is TCIA in the cloud, ready for large-scale model training.