AI trial matching & clinical decision support
Software, increasingly LLM-based, that reads a patient's record and finds trials or guideline options they qualify for.
Only ~5-8% of adult cancer patients enter trials, partly because matching is manual. Tools: TrialGPT (NIH), Tempus TIME, Massive Bio, Deep 6 AI, and hospital-built LLM matchers. Molecular tumour boards use OncoKB and CIViC annotation. Evidence of increased enrolment is emerging.
How it works
Structured data are extracted from the EHR, matched against parsed eligibility criteria, and ranked.
- Scales expert knowledge
- Reduces disparities in trial access
- Hallucination risk; validation
- Eligibility criteria are ambiguous
Latest papers
topQuery for this technology: (TITLE:"clinical trial matching" OR ABSTRACT:"clinical trial matching" OR TITLE:"trial eligibility" OR ABSTRACT:"trial eligibility" OR TITLE:"large language model" OR ABSTRACT:"large language model") AND (cancer OR tumor OR tumour OR oncology OR carcinoma OR lymphoma OR leukemia OR leukaemia OR myeloma OR sarcoma OR melanoma OR glioma). Results are unfiltered search hits about AI trial matching & clinical decision support, not a curated reading list.
Pages like this
not linked directly; found by shared links- TermBasket, umbrella, and platform trials
Shares Just-in-time site activation: open a site in two weeks when a patient appears, N-of-1 and rapid platform trials, Digital twins and virtual control arms, Clinical trial.
- CollectionCIViC
Shares Cancer variant knowledgebases and molecular tumour boards, Clinical NGS bioinformatics and variant interpretation, Every tumour sequencing report lists open, nearby, matched trials pulled live, Knowledge reaches practice too slowly.
- CollectionOncoKB
Shares Cancer variant knowledgebases and molecular tumour boards, Clinical NGS bioinformatics and variant interpretation, Every tumour sequencing report lists open, nearby, matched trials pulled live, Knowledge reaches practice too slowly.
- BottleneckData silos
Shares Direct record-to-database data capture: no manual transcription, no full source verification, Dynamic consent with usage receipts, A live 'seats available' feed for trial slots, like airline inventory, Trial matching inside the electronic record at the moment a treatment is chosen.
- TechnologyDecentralised and hybrid clinical trials
Shares Accrual and enrolment, Clinical trial software (EDC, eCOA, CTMS, RTSM), Unger: most patients never get the chance to join a cancer trial, and when offered, half say yes.
- InstitutionAmerican Society of Clinical Oncology (ASCO)
Shares Just-in-time site activation: open a site in two weeks when a patient appears, HL7 International, Not enough oncologists, nurses, pathologists, physicists, Patients lack understanding, navigation and agency.
- IdeaSet trial enrolment targets from who actually gets the disease, and publish progress live
Shares Unger: most patients never get the chance to join a cancer trial, and when offered, half say yes, ClinicalTrials.gov, Trials do not represent the people who get cancer, Trials enrol too few, too slowly.
- IdeaA cancer data donor card: patient-controlled donation of records for research
Shares Cancer Commons, Patient Data Vault (data-vault.health), Patients lack understanding, navigation and agency.