Questa pagina è stata tradotta automaticamente e l'accuratezza della traduzione non è garantita. Si prega di fare riferimento al Versione inglese per un testo di partenza.

Benchmarking Large Language Models Against Tumour Boards for Oncology Treatment Recommendations (BEACON)

28 luglio 2026 aggiornato da: Assistance Publique - Hôpitaux de Paris

Benchmarking AI for Clinical Oncology decisioNmaking (BEACON): A Prospective, Multicentre, Blinded Evaluation of Frontier Large Language Models Against Multidisciplinary Tumour Board Recommendations in Oncology Treatment Planning

BEACON (Benchmarking AI for Clinical Oncology decisioNmaking) is a prospective, multicentre, comparative, blinded, non-interventional benchmark evaluating the treatment recommendations of five frontier large language models (LLMs) against the recommendations of multidisciplinary tumour boards (RCP) in oncology treatment planning. One hundred standardised synthetic cases (20 per localisation, across breast, lung, urological, digestive and gynaecological cancers) are submitted as identical structured input to two independent tumour boards per localisation and to five frontier LLMs. Each recommendation - human or model - is decomposed into five predefined decision domains (intent, surgery, radiotherapy, systemic therapy, work-up and biomarkers) and scored 0/1/2 for concordance against a two-tier reference: the consensus of the two tumour boards, complemented by an a priori locked guideline matrix (ESMO, NCCN). The primary endpoint is domain-level concordance between LLM and RCP consensus, expressed as a linearly weighted Cohen's kappa. A co-primary safety endpoint captures the proportion of recommendations carrying serious harm potential, because concordance alone can conceal dangerous errors. Because expert boards may disagree with one another on identical cases, model performance is always interpreted against the human consensus. BEACON is designed as reusable, openly licensed, pre-registered infrastructure: all synthetic cases, evaluation rubrics, the locked guideline matrix, scoring algorithms and verbatim prompts are released for full reproducibility.

Panoramica dello studio

Stato

Reclutamento

Condizioni

Intervento / Trattamento

Descrizione dettagliata

BEACON is a prospective, multicentre, blinded benchmark using automated, criteria-based scoring. It is built on three design decisions that distinguish it from the existing literature: (i) synthetic, standardised cases remove the record-completeness variability that confounds retrospective comparisons and allow the identical input to be given to every board and every model; (ii) two independent tumour boards per localisation let human-human agreement be measured rather than assumed; and (iii) a guideline matrix, locked a priori, provides an objective anchor applied identically to human and model recommendations.

Reference standard. For each case-domain, a guideline matrix (guideline-recommended / acceptable / unsupported options per case-domain; ESMO, NCCN), locked and time-stamped before data collection, is applied identically to boards and models.

Five decision domains. Every recommendation is decomposed into D1 Intent, D2 Surgery, D3 Radiotherapy, D4 Systemic therapy (class + line), and D5 Work-up & biomarkers before any comparison.

Tipo di studio

Osservativo

Iscrizione (Stimato)

100

Contatti e Sedi

Questa sezione fornisce i recapiti di coloro che conducono lo studio e informazioni su dove viene condotto lo studio.

Contatto studio

Backup dei contatti dello studio

Luoghi di studio

  • Francia
      • Paris, Francia
        • Reclutamento
        • Hôpital Europeén Georges Pompidou
        • Contatto:

Criteri di partecipazione

I ricercatori cercano persone che corrispondano a una certa descrizione, chiamata criteri di ammissibilità. Alcuni esempi di questi criteri sono le condizioni generali di salute di una persona o trattamenti precedenti.

Criteri di ammissibilità

Età idonea allo studio

  • Adulto
  • Adulto più anziano

Accetta volontari sani

No

Metodo di campionamento

Campione non probabilistico

Popolazione di studio

100 synthetic oncology treatment-planning cases (20 per localisation) across five localisations: breast, lung, urological (prostate, bladder / upper-tract urothelial, kidney), digestive and gynaecological. Each case is a structured JSON input specifying UICC 8th-edition stage, biomarkers, ECOG performance status, comorbidities and a standardised clinical question. No human participants, no patient data and no identifiable individuals. Recommendations are produced by two independent tumour boards per localisation and by five frontier LLMs (queried May 2026).

Descrizione

Inclusion Criteria:

  • Synthetic oncology case within one of the five predefined localisations (breast, lung, urological, digestive, gynaecological).
  • Complete structured schema: UICC 8th-edition stage, biomarkers, ECOG performance status, comorbidities and a standardised clinical question.
  • A clinically answerable treatment-planning question that is mappable to the locked guideline matrix.

Exclusion Criteria:

  • Case outside the five predefined localisations.
  • Incomplete, internally inconsistent or ambiguous schema.
  • Duplicate or near-duplicate of an existing case in the set.
  • Question not resolvable by current guidelines.

Piano di studio

Questa sezione fornisce i dettagli del piano di studio, compreso il modo in cui lo studio è progettato e ciò che lo studio sta misurando.

Come è strutturato lo studio?

Dettagli di progettazione

Coorti e interventi

Gruppo / Coorte
Intervento / Trattamento
Breast cancers
Altro: Multidisciplinary tumour boards
Two independent tumour boards per localisation (10 boards in total) issue a categorical recommendation for every synthetic case. Where both boards agree, their consensus defines the reference standard; where they differ, the case-domain is classified as EQUIPOISE and analysed separately.
Altro: Frontier large language models
Five frontier LLMs (GPT-5.6, Claude Fable 5, Gemini 3.1 Pro, DeepSeek V4 Pro, Llama 4 Maverick) each receive the identical structured input for every case, three times in independent sessions, under locked prompts, versions and settings.
Lung cancers
Altro: Multidisciplinary tumour boards
Two independent tumour boards per localisation (10 boards in total) issue a categorical recommendation for every synthetic case. Where both boards agree, their consensus defines the reference standard; where they differ, the case-domain is classified as EQUIPOISE and analysed separately.
Altro: Frontier large language models
Five frontier LLMs (GPT-5.6, Claude Fable 5, Gemini 3.1 Pro, DeepSeek V4 Pro, Llama 4 Maverick) each receive the identical structured input for every case, three times in independent sessions, under locked prompts, versions and settings.
Urological cancers
Altro: Multidisciplinary tumour boards
Two independent tumour boards per localisation (10 boards in total) issue a categorical recommendation for every synthetic case. Where both boards agree, their consensus defines the reference standard; where they differ, the case-domain is classified as EQUIPOISE and analysed separately.
Altro: Frontier large language models
Five frontier LLMs (GPT-5.6, Claude Fable 5, Gemini 3.1 Pro, DeepSeek V4 Pro, Llama 4 Maverick) each receive the identical structured input for every case, three times in independent sessions, under locked prompts, versions and settings.
Digestive cancers
Altro: Multidisciplinary tumour boards
Two independent tumour boards per localisation (10 boards in total) issue a categorical recommendation for every synthetic case. Where both boards agree, their consensus defines the reference standard; where they differ, the case-domain is classified as EQUIPOISE and analysed separately.
Altro: Frontier large language models
Five frontier LLMs (GPT-5.6, Claude Fable 5, Gemini 3.1 Pro, DeepSeek V4 Pro, Llama 4 Maverick) each receive the identical structured input for every case, three times in independent sessions, under locked prompts, versions and settings.
Gynaecological cancers
Altro: Multidisciplinary tumour boards
Two independent tumour boards per localisation (10 boards in total) issue a categorical recommendation for every synthetic case. Where both boards agree, their consensus defines the reference standard; where they differ, the case-domain is classified as EQUIPOISE and analysed separately.
Altro: Frontier large language models
Five frontier LLMs (GPT-5.6, Claude Fable 5, Gemini 3.1 Pro, DeepSeek V4 Pro, Llama 4 Maverick) each receive the identical structured input for every case, three times in independent sessions, under locked prompts, versions and settings.

Cosa sta misurando lo studio?

Misure di risultato primarie

Misura del risultato
Misura Descrizione
Lasso di tempo
Domain-level performance between LLM recommendations and the locked guidelines.
Lasso di tempo: Assessed once at central scoring, after data collection (~October 2026)
For each recommendation domain and each LLM, proportion of LLM recommendation concordant with locked guidelines
Assessed once at central scoring, after data collection (~October 2026)

Misure di risultato secondarie

Misura del risultato
Misura Descrizione
Lasso di tempo
Proportion of recommendations carrying serious harm potential ( LLM and tumour boards)
Lasso di tempo: Up to October 2026
Up to October 2026
Domain-level recommendation concordance between LLM and tumour-boards
Lasso di tempo: Up to October 2026
Each recommendation domain, decomposed into the five decision domains and scored per domain on an ordinal scale (2 = complete concordance; 1 = partial concordance; 0 = discordance).
Up to October 2026
Inter-tumour board domain-level recommendation concordance
Lasso di tempo: Up to October 2026
Agreement between the two independent tumour boards scored per recommendation domain
Up to October 2026
Equipoise rate
Lasso di tempo: Up to October 2026
Proportion of case-domains where the two tumour boards give different categorical recommendations
Up to October 2026
Completeness
Lasso di tempo: Up to October 2026
Proportion of required domains addressed (LLM and tumour boards)
Up to October 2026
Missingness
Lasso di tempo: Up to October 2026
Proportion of critical omissions (LLM and tumour boards)
Up to October 2026
Intensity bias
Lasso di tempo: Up to October 2026
Proportion of recommendation corresponding to over- or under-treatment
Up to October 2026

Collaboratori e investigatori

Qui è dove troverai le persone e le organizzazioni coinvolte in questo studio.

Sponsor

Assistance Publique - Hôpitaux de Paris

Studiare le date dei record

Queste date tengono traccia dell'avanzamento della registrazione dello studio e dell'invio dei risultati di sintesi a ClinicalTrials.gov. I record degli studi e i risultati riportati vengono esaminati dalla National Library of Medicine (NLM) per assicurarsi che soddisfino specifici standard di controllo della qualità prima di essere pubblicati sul sito Web pubblico.

Studia le date principali

Inizio studio (Effettivo)

1 maggio 2026

Completamento primario (Stimato)

1 ottobre 2026

Completamento dello studio (Stimato)

1 ottobre 2026

Date di iscrizione allo studio

Primo inviato

28 luglio 2026

Primo inviato che soddisfa i criteri di controllo qualità

28 luglio 2026

Primo Inserito (Effettivo)

31 luglio 2026

Aggiornamenti dei record di studio

Ultimo aggiornamento pubblicato (Effettivo)

31 luglio 2026

Ultimo aggiornamento inviato che soddisfa i criteri QC

28 luglio 2026

Ultimo verificato

1 luglio 2026

Maggiori informazioni

Termini relativi a questo studio

Parole chiave

Termini MeSH pertinenti aggiuntivi

Altri numeri di identificazione dello studio

  • APHP261032

Piano per i dati dei singoli partecipanti (IPD)

Hai intenzione di condividere i dati dei singoli partecipanti (IPD)?

INDECISO

Informazioni su farmaci e dispositivi, documenti di studio

Studia un prodotto farmaceutico regolamentato dalla FDA degli Stati Uniti

No

Studia un dispositivo regolamentato dalla FDA degli Stati Uniti

No

Queste informazioni sono state recuperate direttamente dal sito web clinicaltrials.gov senza alcuna modifica. In caso di richieste di modifica, rimozione o aggiornamento dei dettagli dello studio, contattare register@clinicaltrials.gov. Non appena verrà implementata una modifica su clinicaltrials.gov, questa verrà aggiornata automaticamente anche sul nostro sito web .

Sottoscrivi