{"entity":{"id":"ai-oncology-roadmap","kind":"roadmap","name":"AI in oncology roadmap: pattern readers → foundation models → agents in the workflow","aka":[],"tldr":"Artificial intelligence in cancer started as software that flagged spots on a mammogram. It now designs molecules, reads slides better than any single pathologist for some tasks, and is beginning to match patients to trials and draft the tumour board summary; the question is which of it will be proven to help.","summary":"Three strands of AI are converging on oncology. In discovery, structure prediction (AlphaFold 3, Boltz, Chai) and generative chemistry have produced the first AI-designed candidates in trials, and perturbation-scale single-cell datasets are training models that try to predict what a drug will do to a cell. In diagnosis, foundation models trained on millions of slides and scans (Virchow, Prov-GigaPath, UNI, TITAN, CT-FM) underpin the first AI tests cleared to predict treatment benefit (ArteraAI Prostate 2025, ArteraAI Breast 2026) and the first randomised evidence that AI reading improves screening (MASAI). In the clinic, language models are entering trial matching, documentation and tumour-board support, with radiotherapy auto-contouring as the most mature deployed use.\n\nThe gap between the thousands of published models and the handful in clinical use is the defining feature of the field. Prospective, ideally randomised, evidence that an AI-guided decision improves an outcome exists for a few tools; a regulatory route for models that keep updating, payment codes for AI-derived biomarkers, and data that can be shared or federated across hospitals are all unsettled.\n\nThis roadmap covers the whole stack from molecule to clinic; the companion roadmaps go deeper on the AI-assisted clinic and on the virtual cell.","asOf":"2026-09-10","links":[{"label":"MASAI randomised trial of AI-supported mammography screening (Lancet Digital Health 2025)","url":"https://doi.org/10.1016/S2589-7500(24)00267-X"},{"label":"FDA: Artificial intelligence-enabled medical devices list","url":"https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-enabled-medical-devices"}],"tags":[],"related":["ai-oncology-clinic","virtual-cell","idea-multimodal-foundation-model","idea-tr2-ai-external-validation-registry","idea-acc-ai-first-pathology-common-cases","idea-prev-opportunistic-ct-ai-registry","idea-tr1-ai-central-imaging-reads","idea-tr1-ehr-point-of-care-trial-alert","idea-bio1-in-silico-trials-dose","tahoe-100m","arc-virtual-cell-atlas","pathology-benchmarks","imaging-data-commons","flatiron-foundation-cgdb","multidisciplinary-tumour-board","drug-discovery-roadmap"],"cancers":[],"sections":["ai-computation"],"technologies":["radiology-ai-screening","digital-pathology-ai","pathology-foundation-model","ai-drug-design","ai-trial-matching","auto-contouring-ai","federated-learning-medical-ai","digital-twins-trials","ai-compute-platforms","whole-slide-scanners","virchow","prov-gigapath","uni-conch","titan","musk","chief","h-optimus","ct-fm","merlin-ct","aidoc-care","med-gemini","alphafold3","boltz","chai-1","rfdiffusion","esm3","chemistry42","phenom-2","geneformer","scgpt","state-arc","sybil","mirai","foresight-ehr","tempus-multimodal","bionemo"],"targets":[],"drugs":["artera-ai-prostate","artera-ai-breast","paige-prostate"],"companies":["paige","artera","pathai","owkin","tempus","aidoc","isomorphic-labs","insilico-medicine","recursion","xaira-therapeutics","google-deepmind","microsoft-research","nvidia","limbus-ai","therapanacea","trial-library","massive-bio","flatiron-health","pathos-ai","noetik"],"institutions":[],"pathways":[],"terms":[],"trials":["masai"],"people":[],"bottlenecks":["b-ai-validation","b-data-silos","b-real-world-evidence","b-workforce","b-reproducibility"],"keyPapers":[],"journals":[],"dependsOn":[],"notes":[],"steps":[{"era":"1998-2016","title":"Computer-aided detection","description":"The first cleared cancer AI was computer-aided detection for mammography in 1998, which marked suspicious regions for the radiologist and, in large observational studies, did not improve accuracy. Rule-based decision support for treatment recommendations was tried and mostly abandoned. The lesson that survived: an algorithm has to be evaluated on the decision it changes, not on the pattern it finds.","refs":["radiology-ai-screening","mammography"],"status":"historic"},{"era":"2017-2022","title":"Deep learning reaches cleared devices","description":"Convolutional networks trained on labelled images matched specialists on narrow tasks. Paige Prostate (2021) became the first FDA-authorised AI for reading pathology slides; radiology triage tools for haemorrhage and embolism were cleared by the dozen; Sybil and Mirai predicted future lung and breast cancer from today's scan. Whole-slide scanning became routine in large centres, which made slide-level AI possible at all.","refs":["paige-prostate","paige","digital-pathology-ai","whole-slide-scanners","sybil","mirai","aidoc"],"status":"historic"},{"era":"2020-2026","title":"Structure prediction and generative design","description":"AlphaFold made protein structure a lookup rather than a two-year experiment; AlphaFold 3 (2024), Boltz and Chai extended it to drug-protein and antibody complexes, and RFdiffusion and ESM3 design proteins that never existed. Insilico's generative chemistry produced the first AI-discovered drug to reach phase 2, Isomorphic's first oncology candidate was cleared for trials, and Recursion and Xaira are betting that image and perturbation data can find targets no hypothesis would. None has yet produced an approved cancer drug, which is the honest benchmark.","refs":["alphafold3","boltz","chai-1","rfdiffusion","esm3","ai-drug-design","chemistry42","insilico-medicine","isomorphic-labs","recursion","phenom-2","xaira-therapeutics","google-deepmind"],"status":"current"},{"era":"2023-2026","title":"Foundation models and the first predictive tests","description":"Pathology models pretrained on millions of slides (Virchow, Prov-GigaPath, UNI and CONCH, H-optimus, TITAN) predict mutations, biomarkers and outcomes from a routine stain; MUSK adds clinical text. ArteraAI Prostate (2025) was the first AI test cleared to predict benefit from a treatment, and ArteraAI Breast followed in 2026. MASAI gave the first randomised evidence that AI-supported screening finds more cancers with less workload; Aidoc CARE (January 2026) was the first foundation-model triage platform cleared. Single-cell models (Geneformer, scGPT, State) and the Tahoe-100M dataset began the same arc for biology.","refs":["pathology-foundation-model","virchow","prov-gigapath","uni-conch","h-optimus","titan","musk","chief","artera-ai-prostate","artera-ai-breast","artera","masai","aidoc-care","ct-fm","merlin-ct","geneformer","scgpt","state-arc","tahoe-100m","pathology-benchmarks"],"status":"current"},{"era":"2025-2028","title":"Language models enter the workflow","description":"The first widely deployed AI in cancer care is not a diagnosis but a time-saver: auto-contouring of organs and tumours for radiotherapy planning now runs in hundreds of centres. Language models are being tested to match patients to trials from the record at the moment a treatment is chosen, to draft tumour-board summaries and pathology reports, and to answer patient questions under supervision. Federated learning lets models train across hospitals without moving data. The evidence standard for each is still being written.","refs":["auto-contouring-ai","limbus-ai","therapanacea","ai-trial-matching","trial-library","massive-bio","med-gemini","foresight-ehr","federated-learning-medical-ai","owkin","tempus","multidisciplinary-tumour-board","idea-tr1-ehr-point-of-care-trial-alert"],"status":"emerging"},{"era":"2027-2032","title":"From prediction to prospective proof","description":"The field has thousands of retrospective models and a handful of prospective trials. The infrastructure being proposed: a registry of external validation datasets with mandatory reporting, AI-first reading for high-volume common diagnoses with pathologists handling exceptions, every routine CT checked opportunistically for early cancer signs with a tracked pathway, AI central reads to cut trial endpoint cost, and digital twins as virtual control arms where a randomised control is unethical. Regulators are building predetermined change control plans so that models can update without re-clearance.","refs":["idea-tr2-ai-external-validation-registry","idea-acc-ai-first-pathology-common-cases","idea-prev-opportunistic-ct-ai-registry","idea-tr1-ai-central-imaging-reads","digital-twins-trials","imaging-data-commons","flatiron-foundation-cgdb","b-ai-validation"],"status":"emerging"},{"era":"2030+","title":"Patient-level models and the virtual cell","description":"The two long-range bets are a multimodal model that reads slides, scans, genomics and the record to recommend and monitor treatment, and a virtual cell accurate enough to run a drug experiment in silico before it is run in a dish. Both depend on data at a scale no single institution holds, on validation standards that do not yet exist, and on liability and consent questions that are open today. The companion roadmaps on the AI clinic and the virtual cell follow each in detail.","refs":["idea-multimodal-foundation-model","tempus-multimodal","pathos-ai","noetik","arc-virtual-cell-atlas","idea-bio1-in-silico-trials-dose","ai-oncology-clinic","virtual-cell"],"status":"speculative"},{"era":"What sets the pace","title":"Validation, data and compute","description":"The bottleneck is not model quality but the path from a published model to a deployed one: prospective evidence, external validation, regulatory status for updating models, payment, and data that can be shared. Records, scans and genomes sit in silos; real-world outcomes are weakly recorded, so there is little to learn from; and the workforce that would supervise AI is already short. Compute and model platforms are the one input that is not scarce.","refs":["b-ai-validation","b-data-silos","b-real-world-evidence","b-workforce","b-reproducibility","ai-compute-platforms","nvidia","bionemo"],"status":"current"}],"watch":[]},"route":"/roadmaps/ai-oncology-roadmap/","neighbours":{"roadmap":[{"id":"ai-oncology-clinic","kind":"roadmap","name":"AI in the oncology clinic: from narrow cleared tools to multimodal decision support","route":"/roadmaps/ai-oncology-clinic/"},{"id":"drug-discovery-roadmap","kind":"roadmap","name":"Drug discovery roadmap: screening in mice → maps of dependency → designing in silico","route":"/roadmaps/drug-discovery-roadmap/"},{"id":"virtual-cell","kind":"roadmap","name":"Virtual cell roadmap: from bulk omics to a predictive model of a cancer cell","route":"/roadmaps/virtual-cell/"}],"idea":[{"id":"idea-tr2-ai-external-validation-registry","kind":"idea","name":"A registry of external validation datasets for cancer AI models, with mandatory reporting","route":"/ideas/idea-tr2-ai-external-validation-registry/"},{"id":"idea-tr1-ai-central-imaging-reads","kind":"idea","name":"AI-assisted central imaging reads to cut endpoint cost and variability","route":"/ideas/idea-tr1-ai-central-imaging-reads/"},{"id":"idea-acc-ai-first-pathology-common-cases","kind":"idea","name":"AI-first reading for high-volume common cancer diagnoses, pathologist for the exceptions","route":"/ideas/idea-acc-ai-first-pathology-common-cases/"},{"id":"idea-prev-opportunistic-ct-ai-registry","kind":"idea","name":"Every routine CT scan checked by AI for early cancer signs, with a tracked follow-up pathway","route":"/ideas/idea-prev-opportunistic-ct-ai-registry/"},{"id":"idea-bio1-in-silico-trials-dose","kind":"idea","name":"In silico trials to choose the dose before the first patient","route":"/ideas/idea-bio1-in-silico-trials-dose/"},{"id":"idea-multimodal-foundation-model","kind":"idea","name":"Patient-level multimodal foundation models for treatment selection","route":"/ideas/idea-multimodal-foundation-model/"},{"id":"idea-tr1-ehr-point-of-care-trial-alert","kind":"idea","name":"Trial matching inside the electronic record at the moment a treatment is chosen","route":"/ideas/idea-tr1-ehr-point-of-care-trial-alert/"}],"collection":[{"id":"arc-virtual-cell-atlas","kind":"collection","name":"Arc Virtual Cell Atlas","route":"/collections/arc-virtual-cell-atlas/"},{"id":"flatiron-foundation-cgdb","kind":"collection","name":"Flatiron Health and Foundation Medicine Clinico-Genomic Database","route":"/collections/flatiron-foundation-cgdb/"},{"id":"imaging-data-commons","kind":"collection","name":"NCI Imaging Data Commons (IDC)","route":"/collections/imaging-data-commons/"},{"id":"pathology-benchmarks","kind":"collection","name":"Pathology AI benchmarks (CAMELYON, PANDA, TCGA slide tasks)","route":"/collections/pathology-benchmarks/"},{"id":"tahoe-100m","kind":"collection","name":"Tahoe-100M","route":"/collections/tahoe-100m/"}],"technology":[{"id":"auto-contouring-ai","kind":"technology","name":"AI auto-contouring and adaptive planning","route":"/technologies/auto-contouring-ai/"},{"id":"ai-compute-platforms","kind":"technology","name":"AI compute and model platforms for oncology","route":"/technologies/ai-compute-platforms/"},{"id":"radiology-ai-screening","kind":"technology","name":"AI in radiology","route":"/technologies/radiology-ai-screening/"},{"id":"ai-trial-matching","kind":"technology","name":"AI trial matching & clinical decision support","route":"/technologies/ai-trial-matching/"},{"id":"ai-drug-design","kind":"technology","name":"AI-driven drug & target discovery","route":"/technologies/ai-drug-design/"},{"id":"aidoc-care","kind":"technology","name":"Aidoc CARE (clinical radiology foundation model)","route":"/technologies/aidoc-care/"},{"id":"alphafold3","kind":"technology","name":"AlphaFold 3","route":"/technologies/alphafold3/"},{"id":"boltz","kind":"technology","name":"Boltz-1 / Boltz-2 (MIT, open)","route":"/technologies/boltz/"},{"id":"chai-1","kind":"technology","name":"Chai-1 / Chai-2","route":"/technologies/chai-1/"},{"id":"chemistry42","kind":"technology","name":"Chemistry42 and Pharma.AI (Insilico)","route":"/technologies/chemistry42/"},{"id":"chief","kind":"technology","name":"CHIEF (Harvard, Yu Lab)","route":"/technologies/chief/"},{"id":"ct-fm","kind":"technology","name":"CT-FM (whole-body CT foundation model)","route":"/technologies/ct-fm/"},{"id":"digital-pathology-ai","kind":"technology","name":"Digital pathology & AI","route":"/technologies/digital-pathology-ai/"},{"id":"digital-twins-trials","kind":"technology","name":"Digital twins and virtual control arms","route":"/technologies/digital-twins-trials/"},{"id":"esm3","kind":"technology","name":"ESM3 (EvolutionaryScale)","route":"/technologies/esm3/"},{"id":"federated-learning-medical-ai","kind":"technology","name":"Federated learning and privacy-preserving AI","route":"/technologies/federated-learning-medical-ai/"},{"id":"foresight-ehr","kind":"technology","name":"Foresight (generative EHR model)","route":"/technologies/foresight-ehr/"},{"id":"geneformer","kind":"technology","name":"Geneformer","route":"/technologies/geneformer/"},{"id":"h-optimus","kind":"technology","name":"H-optimus (Bioptimus)","route":"/technologies/h-optimus/"},{"id":"mammography","kind":"technology","name":"Mammography & tomosynthesis","route":"/technologies/mammography/"},{"id":"med-gemini","kind":"technology","name":"Med-Gemini and MedLM (Google)","route":"/technologies/med-gemini/"},{"id":"merlin-ct","kind":"technology","name":"Merlin (Stanford abdominal CT vision-language model)","route":"/technologies/merlin-ct/"},{"id":"mirai","kind":"technology","name":"Mirai (MIT breast cancer risk from mammograms)","route":"/technologies/mirai/"},{"id":"multidisciplinary-tumour-board","kind":"technology","name":"Multidisciplinary tumour boards","route":"/technologies/multidisciplinary-tumour-board/"},{"id":"musk","kind":"technology","name":"MUSK (Stanford, vision-language pathology)","route":"/technologies/musk/"},{"id":"bionemo","kind":"technology","name":"NVIDIA BioNeMo","route":"/technologies/bionemo/"},{"id":"pathology-foundation-model","kind":"technology","name":"Pathology & radiology foundation models","route":"/technologies/pathology-foundation-model/"},{"id":"phenom-2","kind":"technology","name":"Phenom-2 and Recursion OS","route":"/technologies/phenom-2/"},{"id":"prov-gigapath","kind":"technology","name":"Prov-GigaPath (Microsoft, Providence)","route":"/technologies/prov-gigapath/"},{"id":"rfdiffusion","kind":"technology","name":"RFdiffusion / RFdiffusion2 and ProteinMPNN (Baker Lab)","route":"/technologies/rfdiffusion/"},{"id":"scgpt","kind":"technology","name":"scGPT","route":"/technologies/scgpt/"},{"id":"state-arc","kind":"technology","name":"State (Arc Institute perturbation model)","route":"/technologies/state-arc/"},{"id":"sybil","kind":"technology","name":"Sybil (MIT/MGH lung cancer risk from CT)","route":"/technologies/sybil/"},{"id":"tempus-multimodal","kind":"technology","name":"Tempus multimodal models","route":"/technologies/tempus-multimodal/"},{"id":"titan","kind":"technology","name":"TITAN (whole-slide multimodal model)","route":"/technologies/titan/"},{"id":"uni-conch","kind":"technology","name":"UNI and CONCH (Harvard, Mahmood Lab)","route":"/technologies/uni-conch/"},{"id":"virchow","kind":"technology","name":"Virchow / Virchow2 (Paige, MSK)","route":"/technologies/virchow/"},{"id":"whole-slide-scanners","kind":"technology","name":"Whole-slide scanners and image management","route":"/technologies/whole-slide-scanners/"}],"section":[{"id":"ai-computation","kind":"section","name":"AI & Computation","route":"/fronts/ai-computation/"}],"drug":[{"id":"artera-ai-breast","kind":"drug","name":"ArteraAI Breast","route":"/drugs/artera-ai-breast/"},{"id":"artera-ai-prostate","kind":"drug","name":"ArteraAI Prostate","route":"/drugs/artera-ai-prostate/"},{"id":"paige-prostate","kind":"drug","name":"Paige Prostate Detect","route":"/drugs/paige-prostate/"}],"company":[{"id":"aidoc","kind":"company","name":"Aidoc","route":"/companies/aidoc/"},{"id":"artera","kind":"company","name":"Artera","route":"/companies/artera/"},{"id":"flatiron-health","kind":"company","name":"Flatiron Health (Roche)","route":"/companies/flatiron-health/"},{"id":"google-deepmind","kind":"company","name":"Google DeepMind (and Google Research)","route":"/companies/google-deepmind/"},{"id":"insilico-medicine","kind":"company","name":"Insilico Medicine","route":"/companies/insilico-medicine/"},{"id":"isomorphic-labs","kind":"company","name":"Isomorphic Labs","route":"/companies/isomorphic-labs/"},{"id":"limbus-ai","kind":"company","name":"Limbus AI","route":"/companies/limbus-ai/"},{"id":"massive-bio","kind":"company","name":"Massive Bio","route":"/companies/massive-bio/"},{"id":"microsoft-research","kind":"company","name":"Microsoft (Research and Health AI)","route":"/companies/microsoft-research/"},{"id":"noetik","kind":"company","name":"Noetik","route":"/companies/noetik/"},{"id":"nvidia","kind":"company","name":"NVIDIA","route":"/companies/nvidia/"},{"id":"owkin","kind":"company","name":"Owkin","route":"/companies/owkin/"},{"id":"paige","kind":"company","name":"Paige AI","route":"/companies/paige/"},{"id":"pathai","kind":"company","name":"PathAI","route":"/companies/pathai/"},{"id":"pathos-ai","kind":"company","name":"Pathos AI","route":"/companies/pathos-ai/"},{"id":"recursion","kind":"company","name":"Recursion Pharmaceuticals","route":"/companies/recursion/"},{"id":"tempus","kind":"company","name":"Tempus AI","route":"/companies/tempus/"},{"id":"therapanacea","kind":"company","name":"TheraPanacea","route":"/companies/therapanacea/"},{"id":"trial-library","kind":"company","name":"Trial Library","route":"/companies/trial-library/"},{"id":"xaira-therapeutics","kind":"company","name":"Xaira Therapeutics","route":"/companies/xaira-therapeutics/"}],"trial":[{"id":"masai","kind":"trial","name":"MASAI (Mammography Screening with Artificial Intelligence)","route":"/trials/masai/"}],"bottleneck":[{"id":"b-ai-validation","kind":"bottleneck","name":"AI that is built but not validated or deployed","route":"/bottlenecks/b-ai-validation/"},{"id":"b-data-silos","kind":"bottleneck","name":"Data silos","route":"/bottlenecks/b-data-silos/"},{"id":"b-workforce","kind":"bottleneck","name":"Not enough oncologists, nurses, pathologists, physicists","route":"/bottlenecks/b-workforce/"},{"id":"b-reproducibility","kind":"bottleneck","name":"Preclinical results do not reproduce","route":"/bottlenecks/b-reproducibility/"},{"id":"b-real-world-evidence","kind":"bottleneck","name":"Weak real-world evidence and registries","route":"/bottlenecks/b-real-world-evidence/"}]}}