n=50

Corpus de fidelidad empírica

La lista pública de los 50 artículos utilizados para medir la fidelidad del scoring (ÉNS, gaussiana neutra H1).

Para medir la fidelidad de cada modo de scoring, usamos un corpus público y estable de 50 artículos temáticos. Distribución alineada con el objetivo teórico (gaussiana neutra, μ=50, σ=20): 5A / 14B / 9C / 14D / 5E + 3 retractados (X).

¿Por qué un corpus fijo? Para que la fidelidad empírica sea reproducible: si cambiamos la rejilla o un evaluador, volvemos a medir sobre el mismo corpus → la brecha es cuantificable, no opinable.

¿Por qué 50 artículos distribución / 28 artículos paired?

De los 50 artículos del corpus, 29 tienen una referencia pdf_ai medible y 28 se utilizan efectivamente para la medición de concordancia (un artículo excluido por resumen demasiado corto). Los otros 21 se puntúan en modo abstract_ai porque su PDF está bloqueado tras el paywall de NEJM, Lancet, JAMA u otras revistas top-tier que no exigen depósito en PMC.

Esta asimetría no es un defecto, es una elección metodológica deliberada que refleja la ciencia médica real: la medicina clínica de vanguardia se publica mayoritariamente en revistas con paywall. Construir un corpus 100% paired habría introducido un sesgo de selección hacia el acceso abierto puro (BMJ, PLOS, Cochrane) y haría inalcanzable una distribución de niveles realista — la mayoría de los artículos del nivel A se publican precisamente en NEJM/Lancet.

Consecuencia operativa: la distribución de niveles (5A/14B/9C/14D/5E/3X) se calcula sobre los 50, y la fidelidad empírica (concordancia abstract_ai vs pdf_ai, concordancia metadata_ai vs pdf_ai) se calcula sobre los 28 paired. Ambas medidas siguen siendo honestas porque responden a dos preguntas diferentes: «¿refleja el corpus la ciencia médica?» y «¿puntúan los modos de forma coherente entre sí?».

Limitación asumida: la medida de concordancia no cubre los artículos del nivel A del corpus (todos abstract-only). Esta limitación es una propiedad estructural del ecosistema editorial científico — no una elección arbitraria nuestra.

Distribución observada vs objetivo H1

ABCDEX
Cible H151491453
Observé51491453

Los 50 artículos del corpus

NivelPuntuaciónPMIDTítuloRevistaAñoModoPairedTemaSeñal
A7836214590Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related DeathThe New England Journal of Medicine2022abstract_aiOnco
A7723117178The benefits and harms of breast cancer screening: an independent review.Lancet (London, England)2012abstract_aiOnco
A7726724178Blood pressure lowering for prevention of cardiovascular disease and death: a systematic review and meta-analysis.Lancet (London, England)2016abstract_aiCardio
A7619297566Screening and prostate-cancer mortality in a randomized European study.The New England Journal of Medicine2009abstract_aiOnco
A7635353979Effect of Early Treatment with Ivermectin among Patients with Covid-19New England Journal of Medicine2022abstract_aiCOVID-iverm
B7337541528Efficacy of Probiotics in Irritable Bowel Syndrome: Systematic Review and Meta-analysisGastroenterology2023abstract_aiMetaEpistemo
B7129897866Primary Prevention of Cardiovascular Disease with a Mediterranean Diet Supplemented with Extra-Virgin Olive Oil or NutsThe New England journal of medicine2018abstract_aiNutrition
B7012421889A population-based study of measles, mumps, and rubella vaccination and autismN Engl J Med2002abstract_aiCOVID-vaccines
B7034002089Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infectionNature medicine2021abstract_aiCOVID-vaccines
B6934145166Ivermectin for Prevention and Treatment of COVID-19 Infection: A Systematic Review, Meta-analysis, and Trial Sequential Analysis to Inform Clinical GuidelinesAmerican Journal of Therapeutics2021abstract_aiCOVID-iverm
B6827717303Ribociclib as First-Line Therapy for HR-Positive, Advanced Breast CancerThe New England Journal of Medicine2016abstract_aiOnco
B6725982160Prognostic value of grip strength: findings from the Prospective Urban Rural Epidemiology (PURE) study.Lancet2015abstract_aiCardio
B649887158Efficacy of bilateral prophylactic mastectomy in women with a family history of breast cancer.The New England journal of medicine1999abstract_aiOnco
B6333567185Once-Weekly Semaglutide in Adults with Overweight or ObesityThe New England Journal of Medicine2021abstract_aiNutrition
B6233545096Azithromycin in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial.Lancet (London, England)2021pdf_aiCOVID-treatments
B6032678530Dexamethasone in Hospitalized Patients with Covid-19.The New England journal of medicine2021pdf_aiCOVID-treatments
B5733378609Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine.The New England journal of medicine2021pdf_aiCOVID-vaccines
B5624519768Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial.BMJ (Clinical research ed.)2014pdf_aiOnco
B5626551272A Randomized Trial of Intensive versus Standard Blood-Pressure Control.The New England journal of medicine2015pdf_aiCardio
C5525073782Fruit and vegetable consumption and mortality from all causes, cardiovascular disease, and cancer: systematic review and dose-response meta-analysis of prospective cohort studies.BMJ (Clinical research ed.)2014pdf_aiNutrition
C5533301246Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine.The New England journal of medicine2020pdf_aiCOVID-vaccines
C5433306989Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK.Lancet (London, England)2021pdf_aiCOVID-vaccines
C5234729549Adverse events of active and placebo groups in SARS-CoV-2 vaccine randomized trials: A systematic review.The Lancet regional health. Europe2022pdf_aiCOVID-vaccines
C5012677558Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case Series unselected for family history: a combined analysis of 22 studies.American journal of human genetics2003pdf_aiGenetics
C4819622552The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration.BMJ (Clinical research ed.)2009pdf_aiMetaEpistemo
C4718997196Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein.The New England Journal of Medicine2008abstract_aiCardio
C4629388196Vaccines for preventing influenza in healthy adultsThe Cochrane Database of Systematic Reviews2018pdf_aiInfectious
C4635854107The serotonin theory of depression: a systematic umbrella review of the evidenceMolecular Psychiatry2022pdf_aiMentalHealth
D4518303940Initial severity and antidepressant benefits: a meta-analysis of data submitted to the Food and Drug Administration.PLoS medicine2008pdf_aiMentalHealth
D4522828485Global trends in antiretroviral resistance in treatment-naive individuals with HIV after rollout of antiretroviral treatment in resource-limited settings: a global collaborative study and meta-regression analysis.Lancet (London, England)2012pdf_aiInfectious
D4535324894Artificial sweeteners and cancer risk: Results from the NutriNet-Santé population-based cohort studyPLOS Medicine2022pdf_aiNutrition
D4420233825Stereotactic body radiation therapy for inoperable early stage lung cancer.JAMA2010pdf_aiOnco
D4426099233Antibiotics for acute otitis media in childrenCochrane Database of Systematic Reviews2015pdf_aiInfectious
D4430305743The UK Biobank resource with deep phenotyping and genomic data.Nature2018pdf_aiGenetics
D4035076665Myocarditis Cases Reported After mRNA-Based COVID-19 Vaccination in the US From December 2020 to August 2021.JAMA2022pdf_aiCOVID-vaccines
D3929477251Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysisThe Lancet2018abstract_aiMentalHealth
D3634477808Surveillance for Adverse Events After COVID-19 mRNA Vaccination.JAMA2021pdf_aiCOVID-vaccines
D3516060722Why most published research findings are false.PLoS medicine2005pdf_aiMetaEpistemo
D3320332511CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials.BMJ (Clinical research ed.)2010pdf_aiMetaEpistemo
D3212813115The Epidemiology of Major Depressive Disorder: Results From the National Comorbidity Survey Replication (NCS-R)JAMA2003abstract_aiMentalHealth
D3018436948GRADE: an emerging consensus on rating quality of evidence and strength of recommendations.BMJ (Clinical research ed.)2008pdf_aiMetaEpistemo
D2920089734Saturated fat, carbohydrate, and cardiovascular disease.The American journal of clinical nutrition2010pdf_aiNutrition
E2517240006B cell mediated priming following pneumococcal colonization.Vaccine2007pdf_aiInfectious
E2529253145Noncommunicable Diseases in People Living With HIV: Time for Integrated CareJ Infect Dis2017abstract_aiInfectious
E2238658043General practice management after transition events: protocol for an experience-based co-design study.BJGP open2024abstract_aiMetaEpistemo
E2240123456Effects of fat emulsion-based early parenteral nutrition for patients after hemihepatectomy.The British journal of nutrition2025abstract_aiNutrition
E2140000001Effect of remineralization product on the microhardness and surface roughness of enamel after bleaching agents.American journal of dentistry2025abstract_aiOther
X032205204RETRACTED: Hydroxychloroquine and azithromycin as a treatment of COVID-19: results of an open-label non-randomized clinical trial.International journal of antimicrobial agents2020pdf_aiCOVID-treatmentsRetractado
X032356626Cardiovascular Disease, Drug Therapy, and Mortality in Covid-19.The New England journal of medicine2020pdf_aiCOVID-treatmentsRetractado
X032450107RETRACTED: Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: a multinational registry analysis.Lancet (London, England)2020pdf_aiCOVID-treatmentsRetractado

Datos brutos

Los 50 JSON están disponibles en el repositorio fuente (`scorings/json/`). Cada artículo contiene el rawScore, los criterios, la integridad y las justificaciones IA. Reproducible $0 mediante `webapp/scripts/replay-fidelity.ts --fidelity-only`.