Immune Super IntelligenceWhere Immunology Meets Advanced AI
Layer 09 · Evidence

Trials as experiments

Each trial is a test of a biological hypothesis in humans. The negative ones teach the most: a mechanism that is real in a dish can be irrelevant, redundant, or protective in a tissue. Claim-level approved support for graph links is still sparse — coverage below is the honest source of truth.

26TRIAL LESSONS
PMIDVERIFIED IDENTIFIERS
Citation curationNeon Auth curation: import a citation manifest, cross-check identifiers against PubMed and Crossref, and approve claim-linked records. Admin is provisioned explicitly.

Evidence coverage

Loading reviewed citations…

CORE GRAPH

0.0%

0 of 395 curated links supported

DRAFT EXTENSIONS

0.0%

0 of 154 draft links supported

SOURCE LINKED, NOT ASSESSED

0

carry a PMID/DOI awaiting claim review

Review status of every link

Supported0An approved source was checked and supports this directional claim.
Partially supported0A reviewed source supports part of this claim.
Conflicting0Reviewed sources disagree about this claim.
Source linked0Carries a PubMed ID or DOI, but nobody has yet checked the source supports this exact claim.
Unreviewed395No source has been mapped to this claim yet.
Draft154Added to close a coverage gap; not yet reviewed by an expert.
Rejected0Reviewed sources do not address this claim.

Supported, by acting concept type

Therapy0/119
Cell0/94
Molecule0/87
Pathway0/65
Disease0/45
Receptor0/38
Trial0/27
Biomarker0/26
Tissue0/17
Discovery0/10
Data0/10
Pathogen0/7
AI method0/4

Most-connected unsupported links

Reviewing these first covers the busiest parts of the graph.

  1. CD8 cytotoxic T lymphocyte drives Acute cellular rejection
  2. CD8 cytotoxic T lymphocyte expresses PD-1
  3. CD8 cytotoxic T lymphocyte responds to SARS-CoV-2
  4. T-cell receptor signalling activates CD8 cytotoxic T lymphocyte
  5. CD8 cytotoxic T lymphocyte kills Tumour cell
  6. B cell differentiates into Plasma cell
  7. Nivolumab rescues from exhaustion CD8 cytotoxic T lymphocyte
  8. PD-L1 suppresses CD8 cytotoxic T lymphocyte
  9. CD8 cytotoxic T lymphocyte differentiates into Exhausted CD8 T cell
  10. CD8 cytotoxic T lymphocyte contributes to Type 1 diabetes

Domain coverage (Phase B)

Educational draft density by thin domain — microbiome, vaccines, transplant, PID, immunogenetics, infection↔autoimmunity, frontier. Gaps are a curator queue, not approved science. Extensions draft-ext-4.

DRAFT CONCEPTS IN DOMAINS

70

7/7 domains have lessons

UNTAUGHT DRAFT CONCEPTS

0

Present in graph but not on a lesson or route

THIN / ISOLATED

0 / 0

Domains below expected min · isolated concepts

Coverage by domain

DomainConceptsLinksTaughtRoutes
1223100%5
1028100%4
1021100%3
1530100%4
1730100%5
515100%2
13100%1

Phase B draft gap fills (coverage + teaching). Uncited draft links in domains: 150 — literature workflow only; never auto-approve.

Knowledge gap queue

Prioritised coverage work. Opening a row inspects the concept or draft link — it does not approve evidence.

  1. Uncited draft link Uncited draft link C015 drives ET02

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  2. Uncited draft link Uncited draft link C018 produces ET04

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  3. Uncited draft link Uncited draft link C018 produces EV04

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  4. Uncited draft link Uncited draft link DS001 characterises EI01

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  5. Uncited draft link Uncited draft link DS004 characterises EF01

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  6. Uncited draft link Uncited draft link EF01 depletes C017

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  7. Uncited draft link Uncited draft link EF01 treats D007

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  8. Uncited draft link Uncited draft link EG01 associates_with EG10

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  9. Uncited draft link Uncited draft link EG02 associates_with D001

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  10. Uncited draft link Uncited draft link EG02 supports B007

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  11. Uncited draft link Uncited draft link EG03 associates_with EG11

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  12. Uncited draft link Uncited draft link EI01 contributes_to D002

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  13. Uncited draft link Uncited draft link EI02 activates C015

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  14. Uncited draft link Uncited draft link EI04 activates C004

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  15. Uncited draft link Uncited draft link EI04 promotes EI02

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  16. Uncited draft link Uncited draft link EM01 produces EM02

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  17. Uncited draft link Uncited draft link EM02 promotes EM03

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  18. Uncited draft link Uncited draft link EM03 expands C014

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  19. Uncited draft link Uncited draft link EM04 triggers P001

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  20. Uncited draft link Uncited draft link EM05 contributes_to D004

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  21. Uncited draft link Uncited draft link EM05 impairs EM02

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  22. Uncited draft link Uncited draft link EP01 depletes C017

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  23. Uncited draft link Uncited draft link EP01 impairs M024

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  24. Uncited draft link Uncited draft link EP02 impairs C018

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  25. Uncited draft link Uncited draft link EP03 impairs M028

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  26. Uncited draft link Uncited draft link EP03 risks X007

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  27. Uncited draft link Uncited draft link EP09 supports C017

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  28. Uncited draft link Uncited draft link ET01 activates C015

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  29. Uncited draft link Uncited draft link ET04 drives ET03

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  30. Uncited draft link Uncited draft link ET04 triggers P013

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  31. Uncited draft link Uncited draft link EV01 activates C004

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  32. Uncited draft link Uncited draft link EV01 drives P008

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  33. Uncited draft link Uncited draft link EV01 generates C026

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  34. Uncited draft link Uncited draft link EV01 triggers P004

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  35. Uncited draft link Uncited draft link EV02 activates C004

    Draft educational edge — queue for literature workflow; do not invent a PMID.

  36. Uncited draft link Uncited draft link R025 supports EG01

    Draft educational edge — queue for literature workflow; do not invent a PMID.

When mechanisms were tested

Bar height is the number of landmark trials in that year. Click a bar to filter.

1999
2000
2003
2006
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2023
1999 → 2023

All landmark trials

2023

CARTITUDE-4positive

HYPOTHESIS · BCMA CAR-T outperforms standard regimens earlier in myeloma.

OUTCOME · Improved progression-free survival vs standard care in lenalidomide-refractory myeloma.

LESSON · Cell therapy moving to earlier lines.

2023

POETYK PSO-1positive

HYPOTHESIS · Selective TYK2 inhibition treats psoriasis.

OUTCOME · Superior to placebo and apremilast.

LESSON · Genetics → allosteric selectivity → approval.

2021

RECOVERYpositive

HYPOTHESIS · Late COVID-19 is driven by the host response.

OUTCOME · Dexamethasone reduced 28-day mortality in patients on oxygen or ventilation; no benefit without oxygen.

LESSON · Same drug, different phase, opposite effect.

2021

NAVIGATORpositive

HYPOTHESIS · TSLP sits upstream of multiple asthma pathways.

OUTCOME · Exacerbations fell, including in low-eosinophil patients.

LESSON · Upstream targets broaden the responsive population.

2020

TULIP-2positive

HYPOTHESIS · Type I IFN drives SLE.

OUTCOME · Anifrolumab met BICLA endpoint (TULIP-1 had missed its primary SRI-4 endpoint).

LESSON · Endpoint choice can decide a trial.

2019

TN10 Teplizumabpositive

HYPOTHESIS · Modulating autoreactive T cells delays T1D onset.

OUTCOME · Median delay ~2 years in stage 2 T1D.

LESSON · Autoimmune disease can be intercepted before symptoms.

2019

COLCOTpositive

HYPOTHESIS · Cheap anti-inflammatory therapy reduces post-MI events.

OUTCOME · Colchicine reduced ischaemic events.

LESSON · Mechanism validated with an old drug.

2019

N-MOmentumpositive

HYPOTHESIS · CD19+ plasmablasts produce pathogenic AQP4-IgG.

OUTCOME · Inebilizumab reduced attacks.

LESSON · Target the antibody-producing cell, not just B cells.

2018

LIBERTY ASTHMA QUESTpositive

HYPOTHESIS · IL-4Rα blockade benefits type-2 asthma.

OUTCOME · Fewer exacerbations; larger benefit with eos ≥300/µL or high FeNO.

LESSON · Biomarkers select the responsive endotype.

2017

CANTOSpositive

HYPOTHESIS · Inflammation independently drives atherothrombosis.

OUTCOME · MACE −15% (HR 0.85) with no LDL change; more fatal infections.

LESSON · Demonstrated the inflammatory hypothesis of atherosclerosis.

2017

ORATORIOpositive

HYPOTHESIS · B cells drive progressive MS.

OUTCOME · Ocrelizumab modestly reduced disability progression in PPMS.

LESSON · First positive progressive MS trial; B cells matter beyond antibody.

2017

TOWERpositive

HYPOTHESIS · Redirecting T cells to CD19 kills B-ALL blasts.

OUTCOME · Longer survival vs chemotherapy.

LESSON · Bispecific engagers turn polyclonal T cells into tumour-specific killers.

2016

LIBERTY ADpositive

HYPOTHESIS · IL-4/IL-13 drive atopic dermatitis.

OUTCOME · Dupilumab improved EASI and itch (SOLO 1/2).

LESSON · Blocking a shared receptor subunit removes two cytokines at once.

2015

KEYNOTE-006positive

HYPOTHESIS · Releasing PD-1 restores anti-tumour CTL function.

OUTCOME · Pembrolizumab beat ipilimumab; ~39% alive at 5 years.

LESSON · Peripheral checkpoints, not priming, limit established tumour immunity.

2015

CheckMate 067positive

HYPOTHESIS · CTLA-4 and PD-1 are non-redundant brakes.

OUTCOME · Combination improved response and survival with more toxicity.

LESSON · Two brakes, two sites, additive efficacy and toxicity.

2014

ERASUREpositive

HYPOTHESIS · IL-17A is the effector of psoriatic plaques.

OUTCOME · Secukinumab produced high PASI 75 rates vs placebo and etanercept.

LESSON · The IL-23/IL-17 axis is the core of psoriasis.

2014

MENSApositive

HYPOTHESIS · Eosinophils drive exacerbations in eosinophilic asthma.

OUTCOME · Mepolizumab halved exacerbations.

LESSON · Earlier anti-IL-5 trials failed because patients were not selected by eosinophils.

2013

GEMINI 1positive

HYPOTHESIS · Blocking gut homing treats UC without systemic immunosuppression.

OUTCOME · Induction and maintenance of remission.

LESSON · Location is a therapeutic dimension.

2012

Secukinumab in Crohn diseasenegative

HYPOTHESIS · IL-17A blockade will help Crohn disease (IL-17 is high in gut).

OUTCOME · Ineffective; more adverse events and worsening.

LESSON · High level ≠ pathogenic role; IL-17A protects the epithelial barrier.

2011

BLISS-52positive

HYPOTHESIS · BAFF sustains autoreactive B cells in SLE.

OUTCOME · Belimumab improved SRI.

LESSON · First new SLE drug in ~50 years.

2010

RAVEpositive

HYPOTHESIS · B cells (ANCA producers) drive vasculitis.

OUTCOME · Rituximab non-inferior to cyclophosphamide; superior in relapsing disease.

LESSON · B-cell depletion can replace alkylating agents.

2006

PNH eculizumab (TRIUMPH)positive

HYPOTHESIS · Terminal complement drives PNH haemolysis.

OUTCOME · Stabilised haemoglobin, reduced transfusion.

LESSON · A single protein target for a single-mechanism disease.

2006

TGN1412 first-in-humansafety signal

HYPOTHESIS · CD28 superagonist will expand Tregs safely.

OUTCOME · Life-threatening cytokine storm in all six volunteers within hours.

LESSON · Human CD4 effector memory T cells express CD28; those of the macaques used for safety testing do not. First-in-human dosing now starts from the minimal anticipated biological effect level (MABEL).

2003

ATTACHnegative

HYPOTHESIS · TNF drives cardiac dysfunction in heart failure.

OUTCOME · High-dose infliximab increased death/hospitalisation.

LESSON · Association-driven targeting can harm; anti-TNF contraindicated in NYHA III–IV.

2000

ATTRACTpositive

HYPOTHESIS · TNF is a causal driver of RA synovitis and erosion.

OUTCOME · Infliximab plus methotrexate slowed radiographic progression.

LESSON · Launched the biologic era; cytokines are druggable.

1999

Lenercept in MSnegative

HYPOTHESIS · Blocking TNF will help MS as it helps RA.

OUTCOME · More frequent and earlier exacerbations.

LESSON · TNFR2 signalling may be protective in the CNS; a cytokine’s role is tissue-specific.