Empirical fidelity corpus
The public list of the 50 publications used to measure the scoring fidelity (ÉNS, neutral Gaussian distribution, μ=50, σ=20).
Current methodology: v2.0.0 (since 10/21/2026)
What is a methodology version?
To measure the fidelity of each scoring mode, we use a public and stable corpus of thematic publications (47 today, target 50). Distribution aligned with the theoretical target (neutral Gaussian, μ=50, σ=20): 5A / 14B / 9C / 14D / 5E + 3 retracted (X) — the observed/target gap is displayed as is, never closed by moving a threshold.
Why a fixed corpus? So that empirical fidelity is reproducible: if we change the grid or an evaluator, we re-measure on the same corpus → the gap is quantifiable, not arguable.
Why 47 publications for the distribution / 34 paired publications?
Of the 47 publications in the corpus, 34 have a measurable Full PDF (AI) reference. The other 13 are scored from the abstract (AI) because their PDF is locked behind the NEJM, Lancet, JAMA, or other major journal paywalls that don't enforce open-access deposit. The published concordance measurement (fidelity badges) was established in May 2026 on 28 paired publications; it will be re-measured at the next replay on the current paired publications.
This asymmetry is not a flaw, it is a deliberate methodological choice that reflects real medical science: cutting-edge clinical medicine is largely published in paywalled journals. Building a 100% paired corpus would have introduced a selection bias toward pure open access (BMJ, PLOS, Cochrane) and made a realistic palier distribution unreachable — most palier A publications are precisely published in NEJM/Lancet.
Operational consequence: the tier distribution (target 5A/14B/9C/14D/5E/3X) is computed on the 47 publications, and the empirical fidelity (Abstract (AI) vs Full PDF concordance, Metadata (Quick) vs Full PDF concordance) on the paired publications. Both measures remain honest because they answer two different questions: 'does the corpus reflect medical science?' and 'do the modes score consistently with each other?'.
Acknowledged limitation: the published concordance measurement (May 2026) included no paired tier-A publication — tier A is rare and largely published in paywalled journals. This limitation is a structural property of the editorial ecosystem — not an arbitrary choice of ours.
Observed distribution vs theoretical target
(ÉNS, neutral Gaussian distribution, μ=50, σ=20)
| A | B | C | D | E | X | |
|---|---|---|---|---|---|---|
| Target distribution | 5 | 14 | 9 | 14 | 5 | 3 |
| Observed | 2 | 14 | 9 | 14 | 5 | 3 |
The 47 publications of the corpus
| Tier | Score | PMID | Title | Journal | Year | Mode | Paired | Theme | Flag |
|---|---|---|---|---|---|---|---|---|---|
| A | 78 | 32445440 | Remdesivir for the Treatment of Covid-19 - Final Report. | The New England journal of medicine | 2020 | Full PDF (AI) | ✓ | COVID-treatments | |
| A | 76 | 33627334 | Statin treatment and muscle symptoms: series of randomised, placebo controlled n-of-1 trials. | BMJ (Clinical research ed.) | 2021 | Full PDF (AI) | ✓ | Cardio | |
| B | 69 | 33205991 | Effectiveness of Adding a Mask Recommendation to Other Public Health Measures to Prevent SARS-CoV-2 Infection in Danish Mask Wearers : A Randomized Controlled Trial. | Annals of Internal Medicine | 2020 | Full PDF (AI) | ✓ | Infectious | |
| B | 68 | 18303940 | Initial severity and antidepressant benefits: a meta-analysis of data submitted to the Food and Drug Administration. | PLOS Medicine | 2008 | Full PDF (AI) | ✓ | MentalHealth | |
| B | 68 | 33882225 | Safety and Efficacy of Single-Dose Ad26.COV2.S Vaccine against Covid-19. | The New England Journal of Medicine | 2021 | Full PDF (AI) | ✓ | COVID-vaccines | |
| B | 66 | 29388196 | Vaccines for preventing influenza in healthy adults. | The Cochrane Database of Systematic Reviews | 2018 | Full PDF (AI) | ✓ | MetaEpistemo | |
| B | 64 | 26099233 | Antibiotics for acute otitis media in children. | Cochrane Database of Systematic Reviews | 2015 | Full PDF (AI) | ✓ | Infectious | |
| B | 64 | 35854107 | The serotonin theory of depression: a systematic umbrella review of the evidence. | Molecular Psychiatry | 2022 | Full PDF (AI) | ✓ | MentalHealth | |
| B | 63 | 40760353 | Ultraprocessed or minimally processed diets following healthy dietary guidelines on weight and cardiometabolic health: a randomized, crossover trial. | Nature Medicine | 2025 | Full PDF (AI) | ✓ | Endocrino | |
| B | 61 | 24519768 | Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial. | BMJ | 2014 | Full PDF (AI) | ✓ | Onco | |
| B | 61 | 31255301 | Population-level impact and herd effects following the introduction of human papillomavirus vaccination programmes: updated systematic review and meta-analysis. | Lancet (London, England) | 2019 | Full PDF (AI) | ✓ | Infectious | |
| B | 60 | 34020944 | Nonreplicable publications are cited more than replicable ones. | Science Advances | 2021 | Full PDF (AI) | ✓ | MetaEpistemo | |
| B | 60 | 36715243 | Physical interventions to interrupt or reduce the spread of respiratory viruses. | The Cochrane database of systematic reviews | 2023 | Full PDF (AI) | ✓ | Infectious | |
| B | 58 | 22998339 | A randomized trial of sugar-sweetened beverages and adolescent body weight. | New England Journal of Medicine | 2012 | Full PDF (AI) | ✓ | Nutrition | |
| B | 58 | 41326666 | Liraglutide in mild to moderate Alzheimer's disease: a phase 2b clinical trial. | Nature Medicine | 2025 | Full PDF (AI) | ✓ | Neurology | |
| B | 56 | 21091279 | Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. | The New England journal of medicine | 2010 | Full PDF (AI) | ✓ | Clinical | |
| C | 54 | 20233825 | Stereotactic body radiation therapy for inoperable early stage lung cancer. | JAMA | 2010 | Full PDF (AI) | ✓ | Onco | |
| C | 52 | 26187233 | A 2-Year Randomized Controlled Trial of Human Caloric Restriction: Feasibility and Effects on Predictors of Health Span and Longevity. | The Journals of Gerontology: Series A | 2015 | Full PDF (AI) | ✓ | Cardio | |
| C | 50 | 40634782 | Plasma proteomics links brain and immune system aging with healthspan and longevity. | Nature Medicine | 2025 | Full PDF (AI) | ✓ | Neurology | |
| C | 49 | 33306989 | Safety 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. | The Lancet | 2021 | Full PDF (AI) | ✓ | Infectious | |
| C | 49 | 34278372 | Impact of original, B.1.1.7, and B.1.351/P.1 SARS-CoV-2 lineages on vaccine effectiveness of two doses of COVID-19 mRNA vaccines: Results from a nationwide case-control study in France. | Lancet Regional Health – Europe | 2021 | Full PDF (AI) | ✓ | Infectious | |
| C | 48 | 25244186 | Ebola virus disease in West Africa--the first 9 months of the epidemic and forward projections. | The New England journal of medicine | 2014 | Full PDF (AI) | ✓ | Infectious | |
| C | 48 | 31621801 | Circulating Tumor DNA Analyses as Markers of Recurrence Risk and Benefit of Adjuvant Therapy for Stage III Colon Cancer. | JAMA oncology | 2019 | Full PDF (AI) | ✓ | Onco | |
| C | 47 | 31922679 | A network meta-analysis of the effects of psychotherapies, pharmacotherapies and their combination in the treatment of adult depression. | World Psychiatry | 2020 | Full PDF (AI) | ✓ | MentalHealth | |
| C | 46 | 40839273 | GLP-1 Receptor Agonists and Cancer Risk in Adults With Obesity. | JAMA Oncology | 2025 | Full PDF (AI) | ✓ | Onco | |
| D | 44 | 41270766 | Ultra-processed foods and human health: the main thesis and the evidence. | Lancet (London, England) | 2025 | Full PDF (AI) | ✓ | Nutrition | |
| D | 43 | 18635428 | Weight loss with a low-carbohydrate, Mediterranean, or low-fat diet. | New England Journal of Medicine | 2008 | Abstract (AI) | — | Nutrition | |
| D | 41 | 16060722 | Why most published research findings are false. | PLoS medicine | 2005 | Full PDF (AI) | ✓ | MetaEpistemo | |
| D | 41 | 32015508 | A new coronavirus associated with human respiratory disease in China. | Nature | 2020 | Full PDF (AI) | ✓ | Infectious | |
| D | 41 | 35116013 | SARS-CoV-2 Vaccination and Protection Against Clinical Disease: A Retrospective Study, Bouches-du-Rhône District, Southern France, 2021. | Frontiers in Microbiology | 2022 | Full PDF (AI) | ✓ | Infectious | |
| D | 40 | 23413192 | Multiple instances of ancient balancing selection shared between humans and chimpanzees. | Science (New York, N.Y.) | 2013 | Full PDF (AI) | ✓ | Clinical | |
| D | 39 | 33835769 | Thrombotic Thrombocytopenia after ChAdOx1 nCov-19 Vaccination. | The New England Journal of Medicine | 2021 | Full PDF (AI) | ✓ | COVID-vaccines | |
| D | 37 | 15328324 | Sugar-sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-aged women. | JAMA | 2004 | Abstract (AI) | — | Nutrition | |
| D | 33 | 31986264 | Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. | Lancet (London, England) | 2020 | Full PDF (AI) | ✓ | Infectious | |
| D | 31 | 10376617 | Autism and measles, mumps, and rubella vaccine: no epidemiological evidence for a causal association. | Lancet | 1999 | Abstract (AI) | — | COVID-vaccines | |
| D | 31 | 23159055 | The Salmonella kinase SteC targets the MAP kinase MEK to regulate the host actin cytoskeleton. | Cell Host Microbe | 2011 | Abstract (AI) | — | Infectious | |
| D | 31 | 28322183 | The UN Decade of Nutrition, the NOVA food classification and the trouble with ultra-processing. | Public Health Nutrition | 2017 | Full PDF (AI) | ✓ | Nutrition | |
| D | 31 | 32459046 | Scientific and Standardization Committee communication: Clinical guidance on the diagnosis, prevention, and treatment of venous thromboembolism in hospitalized patients with COVID-19. | Journal of thrombosis and haemostasis : JTH | 2020 | Full PDF (AI) | ✓ | Infectious | |
| D | 27 | 12421889 | A population-based study of measles, mumps, and rubella vaccination and autism. | N Engl J Med | 2001 | Abstract (AI) | — | COVID-vaccines | |
| E | 25 | 33864435 | How Children and Youth with Medical Complexity Use Hospital and Emergency Department Care across Canada. | Healthcare Quarterly | 2021 | Abstract (AI) | — | Clinical | |
| E | 24 | 10917215 | Serum procalcitonin, C-reactive protein and interleukin-6 for distinguishing bacterial and viral pneumonia in children. | The Pediatric Infectious Disease Journal | 1999 | Abstract (AI) | — | Infectious | |
| E | 24 | 24508041 | Association of the use of MMRV in infants by pediatric infectious disease specialists with that of other affiliated providers. | Vaccine | 2013 | Abstract (AI) | — | COVID-vaccines | |
| E | 23 | 32516120 | [Programming and reprogramming from conception till death]. | New England Journal of Medicine | 2020 | Abstract (AI) | — | Clinical | |
| E | 23 | 38658043 | General practice management after transition events: protocol for an experience-based co-design study. | BJGP open | 2024 | Abstract (AI) | — | MetaEpistemo | |
| X | 0 | 9500320 | Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. | Lancet | 1998 | Abstract (AI) | — | COVID-vaccines | Retracted |
| X | 0 | 23432189 | Primary prevention of cardiovascular disease with a Mediterranean diet. | New England Journal of Medicine | 2013 | Abstract (AI) | — | Nutrition | Retracted |
| X | 0 | 32356626 | Cardiovascular Disease, Drug Therapy, and Mortality in Covid-19. | The New England Journal of Medicine | 2020 | Abstract (AI) | — | Cardio | Retracted |
Column legend
- Mode: method used to score this publication. Full PDF (AI) = full-text analysis; Abstract (AI) = abstract only; Metadata (AI) = title/journal/authors without abstract; Manual entry = manual scoring via form.
- Paired: ✓ if the publication is used in the cross-mode concordance measure (Full PDF reference available). — otherwise (paywalled PDF, measured on its own tier only).
Raw data
The 47 scoring files are publicly available in the open source repository. Each file contains, for one publication, its detailed score, the per-criterion breakdown, the integrity coefficient and the AI justifications. The fidelity measurement is fully reproducible and free from these files.