Nuclide

Roadmap

Every upgrade idea we have, with a status the corpus can contradict. The gauges are computed from the data at build time; where a claim of “shipped” is not borne out, the idea is shown as needing work and the gauge says what to do.

A. Corpus health

23 coverage checks over 862 records, recomputed on every build from src/lib/health.ts. Each gauge is passing records over records checked; each names the worst offenders and the one edit that fixes them. Adding a check is one entry in an array.

10 met13 need workSorted worst first. The tick on each bar is the target; a “shipped” idea below its target is shown as “needs work” in the list below.
0/ 140%

Each bottleneck should have at least ten ideas attacking it; fewer means the ideas board has not reached it yet.

What to do

Add idea records under data/ideas-waves/ with this bottleneck id in `bottlenecks`.

All 14 failing
  1. Actinium-225 supply0 ideas
  2. Authorized user shortage for radionuclide therapy0 ideas
  3. Clinical trial capacity for radioligand therapy0 ideas
  4. Cold chain and time-critical logistics0 ideas
  5. Dosimetry standardisation for radioligand therapy0 ideas
  6. Enriched target material for therapeutic isotopes0 ideas
  7. Hot lab and shielded treatment room capacity0 ideas
  8. Molybdenum-99 supply0 ideas
  9. Nuclear medicine technologist and physicist workforce0 ideas
  10. Radioactive waste handling and disposal regulation0 ideas
  11. Reimbursement and coding for radiopharmaceuticals0 ideas
  12. Site readiness and referral pathway delays0 ideas
  13. Wrong doses0 ideas
  14. Manufacturing cost and time for living and radioactive medicines8 ideas
67/ 8627.8%

Every page should show its last commit, author, and diff; that comes from public/provenance.json, rebuilt weekly.

What to do

Run `npm run provenance` (the weekly fact-check workflow does this) so new records get a provenance line.

Worst 25 of 795 failing
  1. HER2-positive breast cancerno provenance entry
  2. Diffuse large B-cell lymphomano provenance entry
  3. Follicular lymphomano provenance entry
  4. Glioma & glioblastomano provenance entry
  5. Grade 3 well-differentiated neuroendocrine tumourno provenance entry
  6. Hepatocellular carcinomano provenance entry
  7. Intermediate hepatocellular carcinoma (BCLC B)no provenance entry
  8. Hereditary pheochromocytoma and paraganglioma (SDHx, VHL, RET, NF1, MAX and TMEM127)no provenance entry
  9. Lung neuroendocrine tumours (typical and atypical carcinoid)no provenance entry
  10. Meningiomano provenance entry
  11. Metastatic pheochromocytoma and paragangliomano provenance entry
  12. Multiple endocrine neoplasia syndromes (MEN1, MEN2, MEN4)no provenance entry
  13. Neuroblastoma (paediatric)no provenance entry
  14. High-risk neuroblastomano provenance entry
  15. Neuroendocrine tumoursno provenance entry
  16. Osteosarcomano provenance entry
  17. Pancreatic neuroendocrine tumoursno provenance entry
  18. Pheochromocytoma and paraganglioma (PPGL)no provenance entry
  19. Biochemical recurrence of prostate cancerno provenance entry
  20. Localised prostate cancer, high and very high riskno provenance entry
  21. Metastatic castration-resistant prostate cancerno provenance entry
  22. Metastatic hormone-sensitive prostate cancerno provenance entry
  23. Neuroendocrine and small-cell prostate cancerno provenance entry
  24. Non-metastatic castration-resistant prostate cancerno provenance entry
  25. Renal cell carcinomano provenance entry
5/ 1623.1%

Companies, institutions, and collections should show a self-hosted logo so lists are scannable.

What to do

Run `npm run fetch:logos`; add a Wikidata QID override in the script if the automatic match fails.

Worst 25 of 157 failing
  1. Aktis Oncologyno logo
  2. Alpha-9 Oncologyno logo
  3. ANSTO (Australian Nuclear Science and Technology Organisation)no logo
  4. Ariceum Therapeuticsno logo
  5. ARTBIOno logo
  6. AstraZenecano logo
  7. Atomic Alchemyno logo
  8. Bayerno logo
  9. Blue Earth Diagnostics (Bracco)no logo
  10. Bristol Myers Squibbno logo
  11. Boston Scientificno logo
  12. Canon Medical Systemsno logo
  13. Cardinal Health Nuclear & Precision Health Solutionsno logo
  14. Cellectar Biosciencesno logo
  15. Clarity Pharmaceuticalsno logo
  16. Comecer (ATS)no logo
  17. Convergent Therapeuticsno logo
  18. Curiumno logo
  19. Eckert & Zieglerno logo
  20. Eli Lilly (incl. Loxo)no logo
  21. Evergreen Theragnosticsno logo
  22. Fusion Pharmaceuticals (AstraZeneca)no logo
  23. FutureChemno logo
  24. GE HealthCareno logo
  25. Hamamatsu Photonicsno logo
11/ 3630.6%

For each target we should say how common it is in each cancer, with a source, so the prevalence matrix is complete.

What to do

Add `prevalence` rows ({ indicationId, pct, measure, source }) to the target record.

All 25 failing
  1. 5T4 (trophoblast glycoprotein)no prevalence rows
  2. Alpha-synucleinno prevalence rows
  3. Amyloid beta plaquesno prevalence rows
  4. Bone mineral (hydroxyapatite)no prevalence rows
  5. Bradykinin B1 receptor (B1R)no prevalence rows
  6. CAIX (carbonic anhydrase IX)no prevalence rows
  7. CCK2 receptor (gastrin receptor)no prevalence rows
  8. CEACAM5no prevalence rows
  9. CXCR4no prevalence rows
  10. DLG4 (PSD-95)no prevalence rows
  11. Dopamine transporter (DAT)no prevalence rows
  12. FAPno prevalence rows
  13. Folate receptor alphano prevalence rows
  14. GLUT1 / hexokinase (glucose metabolism)no prevalence rows
  15. Integrin αvβ3no prevalence rows
  16. KLK2 (kallikrein-2)no prevalence rows
  17. Mesothelin (MSLN)no prevalence rows
  18. Nectin-4no prevalence rows
  19. NIS (sodium-iodide symporter)no prevalence rows
  20. Norepinephrine transporter (NET)no prevalence rows
  21. Sortilin (SORT1)no prevalence rows
  22. STEAP1no prevalence rows
  23. Tau protein (neurofibrillary tangles)no prevalence rows
  24. TROP2no prevalence rows
  25. uPAR (urokinase plasminogen activator receptor)no prevalence rows
31/ 7640.8%

Glossary terms promise a Wikipedia link for readers who want the long version.

What to do

Set `wikipedia` on the term record to the matching article URL.

Worst 25 of 45 failing
  1. Absorbed dose and the gray (Gy)no Wikipedia link
  2. Activity prescription: fixed versus dosimetry-guided dosingno Wikipedia link
  3. ALARA (as low as is reasonably achievable)no Wikipedia link
  4. Attenuation correctionno Wikipedia link
  5. Authorized User (AU)no Wikipedia link
  6. Becquerel and curie (units of radioactivity)no Wikipedia link
  7. Biological half-lifeno Wikipedia link
  8. Chelator (bifunctional chelating agent)no Wikipedia link
  9. Deauville five-point scale for PETno Wikipedia link
  10. Decay correctionno Wikipedia link
  11. Dose calibratorno Wikipedia link
  12. Effective half-lifeno Wikipedia link
  13. Equivalent dose and effective dose (sievert)no Wikipedia link
  14. Fixed versus personalised (dosimetry-guided) dosingno Wikipedia link
  15. Flare phenomenonno Wikipedia link
  16. Hot lab (radiopharmacy dispensing area)no Wikipedia link
  17. Krenning scoreno Wikipedia link
  18. Linear energy transfer (LET)no Wikipedia link
  19. Medicare coverage with evidence developmentno Wikipedia link
  20. MIBG Curie and SIOPEN scoresno Wikipedia link
  21. MIRD formalismno Wikipedia link
  22. Myelosuppression after radioligand therapyno Wikipedia link
  23. No-carrier-added (n.c.a.) and carrier-added productionno Wikipedia link
  24. NRC medical-use licensing (10 CFR Part 35)no Wikipedia link
  25. Oligometastatic diseaseno Wikipedia link

Institutions with peopleinstitutions

needs work
19/ 6031.7%

An institution page should name the clinicians and scientists who work there.

What to do

Add person records under data/people/ with `institutionId` set to this institution.

Worst 25 of 41 failing
  1. All India Institute of Medical Sciences, New Delhino people
  2. American Society of Nuclear Cardiologyno people
  3. Board of Radiation and Isotope Technologyno people
  4. Centre Eugène Marquisno people
  5. Charité Universitätsmedizin Berlinno people
  6. Chinese PLA General Hospitalno people
  7. Chinese Society of Nuclear Medicineno people
  8. Comprehensive Cancer Center Mainfranken, University Hospital Würzburgno people
  9. Dana-Farber Brigham Cancer Centerno people
  10. Erasmus MC Cancer Instituteno people
  11. European Medicines Agencyno people
  12. Fundación Instituto Valenciano de Oncología (IVO)no people
  13. Hospital Universitari i Politècnic La Feno people
  14. Inselspital, Bern University Hospital / University Cancer Center Inselspitalno people
  15. Institut de Cancérologie de l'Ouest (ICO)no people
  16. Institut Jules Bordetno people
  17. IRCCS Istituto Romagnolo per lo Studio dei Tumori 'Dino Amadori' (IRST)no people
  18. IRCCS Sacro Cuore Don Calabria Hospitalno people
  19. Istituto Oncologico Veneto IRCCSno people
  20. Japanese Society of Nuclear Medicineno people
  21. Johns Hopkins Hospital / Sidney Kimmel Comprehensive Cancer Centerno people
  22. Karolinska University Hospitalno people
  23. King Faisal Specialist Hospital and Research Centreno people
  24. Korea Institute of Radiological and Medical Sciencesno people
  25. Korean Society of Nuclear Medicineno people
24/ 6934.8%

Each technology page should show a wireframe schematic of its mechanism, its own or an aliased drawing of the same process, not the generic one borrowed from its front.

What to do

Add a builder for this id in data/schematics.ts (static) or data/animated.ts (animated).

Worst 25 of 45 failing
  1. AI in nuclear medicinegeneric schematic
  2. Albumin-binder ligandsgeneric schematic
  3. Alpha-emitter daughter redistribution ('free daughter' problem)generic schematic
  4. Alpha-emitter nanogenerators and daughter trappinggeneric schematic
  5. Auger-electron therapygeneric schematic
  6. Automated synthesis modulesgeneric schematic
  7. CAIX PET (89Zr-girentuximab)generic schematic
  8. Cold kit formulationgeneric schematic
  9. Cyclotron isotope production (F-18, Ga-68, Cu-64, Zr-89, At-211)generic schematic
  10. DOTA chelationgeneric schematic
  11. FES PET (oestrogen receptor imaging)generic schematic
  12. Gamma probes, handheld gamma cameras and dose calibratorsgeneric schematic
  13. Germanium-68/gallium-68 generatorsgeneric schematic
  14. Hot cells and radiation shieldinggeneric schematic
  15. In vivo dosimetry and patient-specific quality assurancegeneric schematic
  16. Low-dose-rate seed brachytherapygeneric schematic
  17. Macropa chelationgeneric schematic
  18. Medical cyclotrons, hot cells, and synthesis modulesgeneric schematic
  19. MIBG imaging and 131I-MIBG therapygeneric schematic
  20. Molybdenum-99/technetium-99m generatorsgeneric schematic
  21. Monte Carlo dosimetrygeneric schematic
  22. Nuclear medicine and total-body PET hardwaregeneric schematic
  23. Peptide receptor radionuclide therapy (PRRT)generic schematic
  24. PET tracer manufacturing and distributiongeneric schematic
  25. PET-adapted (response-adapted) therapygeneric schematic

Indications with a deep diveindications

needs work
32/ 5162.7%

Every cancer page should reach TNBC depth: at least three standard-of-care settings, three history events, and a pipeline.

What to do

Add `standardOfCare` rows by setting with refs, `history` events with refs, and `pipeline` ids; see indications.ts for the TNBC model.

All 19 failing
  1. Alzheimer's disease0 standard-of-care rows, 0 history events, empty pipeline
  2. Parkinsonian syndromes0 standard-of-care rows, 0 history events, empty pipeline
  3. Cardiac amyloidosis (transthyretin, ATTR-CM)1 standard-of-care rows, 0 history events, empty pipeline
  4. Cardiac sarcoidosis1 standard-of-care rows, 0 history events, empty pipeline
  5. Coronary artery disease1 standard-of-care rows, 0 history events, empty pipeline
  6. Fever of unknown origin2 standard-of-care rows, 0 history events, empty pipeline
  7. Gastrointestinal bleeding localisation2 standard-of-care rows, 0 history events, empty pipeline
  8. Presurgical localisation of drug-resistant focal epilepsy2 standard-of-care rows, 0 history events, empty pipeline
  9. Pulmonary embolism: ventilation/perfusion imaging2 standard-of-care rows, 0 history events, empty pipeline
  10. Sentinel lymph node mapping in breast cancer and melanoma2 standard-of-care rows, 0 history events, empty pipeline
  11. Bone pain palliation from osteoblastic metastases0 history events, empty pipeline
  12. Functional assessment of thyroid nodules0 history events, empty pipeline
  13. Graves' disease and hyperthyroidism0 history events, empty pipeline
  14. Prosthetic joint infection0 history events, empty pipeline
  15. Suspected upper urinary tract obstruction0 history events, empty pipeline
  16. Neuroendocrine and small-cell prostate cancerempty pipeline
  17. Follicular lymphomaempty pipeline
  18. Glioma & glioblastomaempty pipeline
  19. Diffuse large B-cell lymphomaempty pipeline
64/ 10064%

Small molecules should rotate on their page; biologics, cells, and tests should show a placeholder that says why there is no molecule.

What to do

Add the product to structures.ts (PubChem CID or PDB id) and run `npm run fetch:structures`, or set a `modality` the placeholder recognises.

Worst 25 of 36 failing
  1. Abiraterone acetatemodality "Small-molecule CYP17A1 inhibitor" has no structure or placeholder
  2. Ammonia N-13modality "PET tracer" has no structure or placeholder
  3. Cabazitaxelmodality "Cytotoxic chemotherapy (taxane)" has no structure or placeholder
  4. Cabozantinibmodality "Small-molecule multi-kinase inhibitor (MET, VEGFR2, AXL, RET)" has no structure or placeholder
  5. Capecitabinemodality "Oral fluoropyrimidine prodrug (antimetabolite)" has no structure or placeholder
  6. Choline C-11modality "PET radiotracer, intravenous" has no structure or placeholder
  7. Copper-64 SAR-bisPSMAmodality "PET tracer" has no structure or placeholder
  8. Dexrazoxanemodality "Cardioprotectant and extravasation antidote (bisdioxopiperazine)" has no structure or placeholder
  9. Docetaxelmodality "Cytotoxic chemotherapy (taxane)" has no structure or placeholder
  10. Donafenibmodality "Small-molecule multi-kinase inhibitor (deuterated sorafenib)" has no structure or placeholder
  11. Enzalutamidemodality "Small-molecule AR antagonist" has no structure or placeholder
  12. Everolimusmodality "Small-molecule mTOR inhibitor" has no structure or placeholder
  13. Florastamin F-18modality "PSMA-targeted PET radiotracer, fluorine-18" has no structure or placeholder
  14. Florquinitau F-18modality "PET tracer" has no structure or placeholder
  15. Gallium citrate Ga-67modality "SPECT tracer" has no structure or placeholder
  16. Gallium-68 FAPI-46modality "PET tracer" has no structure or placeholder
  17. Gallium-68 pentixaformodality "PET tracer" has no structure or placeholder
  18. Holmium-166 microspheresmodality "radioembolisation" has no structure or placeholder
  19. HRS-4357modality "not stated (registered as a drug or biological on ClinicalTrials.gov)" has no structure or placeholder
  20. Iodine-125 brachytherapy seedsmodality "brachytherapy source" has no structure or placeholder
  21. Iodine-131 tositumomabmodality "radioimmunotherapy" has no structure or placeholder
  22. Lutetium-177 rosopatamab tetraxetanmodality "radioimmunotherapy" has no structure or placeholder
  23. Olaparibmodality "Small-molecule PARP inhibitor" has no structure or placeholder
  24. Pasireotidemodality "Multireceptor somatostatin analogue (subcutaneous or monthly depot)" has no structure or placeholder
  25. Rhenium-188 skin cancer therapymodality "brachytherapy source" has no structure or placeholder
655/ 85077.1%

A record nothing links to can only be found by search; it should be referenced by at least one other page.

What to do

Find the cancer, product, trial, or idea this record belongs with and add this id to its relation arrays.

Worst 25 of 195 failing
  1. Scandium-47no incoming links
  2. Multiple endocrine neoplasia syndromes (MEN1, MEN2, MEN4)no incoming links
  3. Sentinel lymph node mapping in breast cancer and melanomano incoming links
  4. Suspected upper urinary tract obstructionno incoming links
  5. Gastrointestinal bleeding localisationno incoming links
  6. Bone pain palliation from osteoblastic metastasesno incoming links
  7. Presurgical localisation of drug-resistant focal epilepsyno incoming links
  8. Donafenibno incoming links
  9. Gallium citrate Ga-67no incoming links
  10. Florquinitau F-18no incoming links
  11. Iodine-131 tositumomabno incoming links
  12. Strontium-89 chlorideno incoming links
  13. Rhenium-188 skin cancer therapyno incoming links
  14. Lutetium-177 PNT2002no incoming links
  15. Lutetium-177 rosopatamab tetraxetanno incoming links
  16. Lutetium-177 dotatate (radioligand equivalent)no incoming links
  17. Iodine-125 brachytherapy seedsno incoming links
  18. Actinium-225 FPI-2068no incoming links
  19. Folate receptor alphano incoming links
  20. NIS (sodium-iodide symporter)no incoming links
  21. Norepinephrine transporter (NET)no incoming links
  22. Bone mineral (hydroxyapatite)no incoming links
  23. CCK2 receptor (gastrin receptor)no incoming links
  24. uPAR (urokinase plasminogen activator receptor)no incoming links
  25. Mesothelin (MSLN)no incoming links
0/ 8500%

Pages should carry a named expert or patient-advocate reviewer with a date and a conflict-of-interest statement.

What to do

Recruit a reviewer for this record and add an entry to data/reviews.ts (track, reviewer, role, date, coi).

Worst 25 of 850 failing
  1. HER2-positive breast cancernot reviewed
  2. Diffuse large B-cell lymphomanot reviewed
  3. Follicular lymphomanot reviewed
  4. Glioma & glioblastomanot reviewed
  5. Grade 3 well-differentiated neuroendocrine tumournot reviewed
  6. Hepatocellular carcinomanot reviewed
  7. Intermediate hepatocellular carcinoma (BCLC B)not reviewed
  8. Hereditary pheochromocytoma and paraganglioma (SDHx, VHL, RET, NF1, MAX and TMEM127)not reviewed
  9. Lung neuroendocrine tumours (typical and atypical carcinoid)not reviewed
  10. Meningiomanot reviewed
  11. Metastatic pheochromocytoma and paragangliomanot reviewed
  12. Multiple endocrine neoplasia syndromes (MEN1, MEN2, MEN4)not reviewed
  13. Neuroblastoma (paediatric)not reviewed
  14. High-risk neuroblastomanot reviewed
  15. Neuroendocrine tumoursnot reviewed
  16. Osteosarcomanot reviewed
  17. Pancreatic neuroendocrine tumoursnot reviewed
  18. Pheochromocytoma and paraganglioma (PPGL)not reviewed
  19. Prostate cancernot reviewed
  20. Biochemical recurrence of prostate cancernot reviewed
  21. Localised prostate cancer, high and very high risknot reviewed
  22. Metastatic castration-resistant prostate cancernot reviewed
  23. Metastatic hormone-sensitive prostate cancernot reviewed
  24. Neuroendocrine and small-cell prostate cancernot reviewed
  25. Non-metastatic castration-resistant prostate cancernot reviewed
52/ 5496.3%

An approved product should say where it is approved (US, EU, UK, Japan, China, Australia), not just that it is; at minimum the FDA or EMA verdict, unless the product is approved only elsewhere.

What to do

Add a row for this product in regional-approvals.ts with a regulator source per region (US or EU at least).

All 2 failing
  1. Iodine-125 brachytherapy seedsno regional row
  2. Rhenium-188 skin cancer therapyno regional row
856/ 86299.3%

Every record should cite at least one external source (a label, paper, registry, or regulator page) so a reader can check it.

What to do

Add a `links` entry with a primary source URL and a label. For trials an `nct` id counts.

All 6 failing
  1. TROP2 PETno source
  2. Chris C. Parkerno source
  3. MIBG Curie and SIOPEN scoresno source
  4. PSA kinetics: PSA doubling time and PSA densityno source
  5. SDH deficiency (SDHB immunohistochemistry loss)no source
  6. Alpha radioligands after ADC failureno source
856/ 86299.3%

The technical summary should say what the thing is, how it works, and what the evidence shows; under 300 characters it cannot.

What to do

Expand `summary` with mechanism, evidence, and open questions, in paragraphs separated by blank lines.

All 6 failing
  1. Excel Diagnostics and Nuclear Oncology Center219 characters
  2. PSMA-PET to Guide Prostatectomy227 characters
  3. HER2 Targeted Molecular Imaging in mBC and Other Metastatic Solid Carcinomas Using 68Ga-ABS011273 characters
  4. Test-retest Evaluation of [18F]F-AraG PET279 characters
  5. A Trial of HRS-6768 in Patients With Advanced Solid Tumors294 characters
  6. A Phase 1/2 Study of 177Lu-NYM032 Injection in mCRPC298 characters
23/ 23100%

Each key paper should show how often it has been cited, from the Europe PMC index; a paper with no DOI or PMID cannot be looked up at all.

What to do

If the paper has no DOI or PMID, add one to its record; then run `npm run fetch:citations` (weekly workflow, budget-free) so the paper appears in public/citations/index.json.

B. Ideas by status

All 45 product, data, and community ideas from three waves, grouped by theme. The number is wave and position (W2·07). Editorial statuses are shipped, building, planned, proposed; needs work is applied automatically when a gauge linked to the idea is below its target, which is currently the case for 11 ideas previously marked shipped or building.

11 needs work4 building6 planned5 proposed19 shipped

Knowledge graph 10

  1. W1·01
    Page per object, backlinks everywhere

    Every isotope, agent, target, trial, company, institution and term has a URL and shows what links to it. That is what makes a hub rather than a list.

    Editorial status was “shipped”, but records something links to is 655/850 (77.1%, target 95%). Claim it.

    needs work
  2. W1·02
    Isotope as a first-class object

    Radiopharmacy starts with the nuclide: half-life, emission, tissue range, how it is made and whether anyone can get it. Every isotope is a record, and agents link to the isotope they carry.

    Editorial status was “shipped”, but kinds with 25 or more records is 11/20 (55%, target 95%). Claim it.

    needs work
  3. W3·43
    Deep dives per indication

    For the indications where radiopharmaceuticals matter most — prostate, NET, thyroid, glioma, bone metastases — a longer state-of-the-art page rather than a stub.

    Editorial status was “building”, but indications with a deep dive is 32/51 (62.7%, target 95%). Claim it.

    needs work
  4. W1·09
    Wikidata and Wikipedia cross-linking

    Push structured nuclide and agent facts to Wikidata, pull summaries back, so the hub and the commons improve each other.

    Verified: terms with a wikipedia link is 31/76 (40.8%, target 95%).

    proposed
  5. W1·03
    Plain-English TL;DR on every page

    Patients, families, journalists and investors should get the point in one sentence before the physics.

    Verified: tl;drs that are full sentences is 862/862 (100%, target 95%).

    shipped
  6. W1·04
    Public JSON API of the whole corpus

    Let others build on the data: dose calculators, trial matchers, supply dashboards. Published as static JSON under /api/v1/.

    shipped
  7. W1·05
    Graph explorer

    Navigate visually from an isotope to its agents to their targets, trials and manufacturers.

    shipped
  8. W1·06
    'As of' dates and a change log on every fact

    Approvals, supply and trial status move monthly. Show when each fact was checked and what changed, like a package changelog.

    Verified: records checked in the last 60 days is 862/862 (100%, target 90%).

    shipped
  9. W1·07
    Evidence tiers as a visual language

    Colour and badge every claim by evidence level — approved, phase 3, phase 2, first-in-human, preclinical, concept — so hype is visible at a glance.

    shipped
  10. W1·08
    A primary source on every record

    A label, approval letter, EPAR, registry entry, guideline or paper for each record, so a reader can check the claim rather than trust us.

    Verified: records with a primary source is 856/862 (99.3%, target 95%).

    shipped

Theranostics 5

  1. W1·13
    Alpha-emitter tracker

    225Ac, 211At, 212Pb, 227Th and 223Ra each have their own supply, chemistry and recoil problems. One page per nuclide and one comparison across them.

    building
  2. W1·11
    Target-by-isotope matrix

    One grid of what has been tried against each target with each nuclide — PSMA with 177Lu, 225Ac, 161Tb — and where the gaps are.

    planned
  3. W1·12
    Dosimetry method registry

    Trials quote absorbed dose in incompatible ways. Record the method (SPECT/CT time points, planar, hybrid, software) per trial so numbers can be compared honestly.

    Verified: trials with structured outcomes is 38/39 (97.4%, target 95%).

    planned
  4. W1·10
    Theranostic pairs as a first-class relation

    The field's central idea is see-and-treat: 68Ga with 177Lu, 64Cu with 67Cu, 124I with 131I. Pairs are recorded on the isotope and shown on both pages.

    shipped
  5. W1·14
    Pairings and anti-pairings

    Combinations that work (radioligand plus PARP inhibitor, plus immune checkpoint), sequences that work, and cautions, each with its rationale and evidence.

    shipped

Diagnostics 4

  1. W1·17
    Scanner and detector technology pages

    Long-axial-field-of-view PET, CZT SPECT, total-body imaging and dose-reduction reconstruction decide what is clinically possible; they belong in the graph beside the tracers.

    Editorial status was “shipped”, but technologies with a specific schematic is 24/69 (34.8%, target 80%). Claim it.

    needs work
  2. W1·16
    Reporting-criteria pages (PSMA-RADS, Krenning, Deauville, PERCIST)

    A scan means nothing without the criteria used to read it. Each criterion gets a glossary record linked from the agents that use it.

    building
  3. W1·15
    Diagnostic agents treated as equals, not footnotes

    PET and SPECT tracers are the larger half of nuclear medicine by patient numbers. Every approved tracer gets a record with its indication, label and evidence.

    Verified: approved products with regional rows is 52/54 (96.3%, target 95%).

    shipped
  4. W1·18
    Beyond oncology: cardiac, neurology, infection, endocrine

    Amyloid and tau PET, myocardial perfusion and amyloidosis tracers, infection imaging and thyroid therapy are nuclear medicine too, and most radiopharma coverage ignores them.

    shipped

Supply 5

  1. W1·21
    Manufacturer and CDMO directory

    Who can label, fill and release a clinical batch, in which country, at what activity — the question every programme asks in year two.

    Editorial status was “shipped”, but organisations with a logo is 5/162 (3.1%, target 90%). Claim it.

    needs work
  2. W1·22
    Bottleneck pages with named ideas against them

    Each bottleneck (225Ac supply, chelator patents, dosimetry standards, scanner time, authorised-user shortage) gets a page listing what could unblock it.

    Editorial status was “shipped”, but bottlenecks with ten or more ideas is 0/14 (0%, target 95%). Claim it.

    needs work
  3. W1·20
    Half-life-aware logistics explainer

    A 68-minute nuclide cannot be shipped like a 6.6-day one. Make the distribution radius and the cold-chain-free reality explicit on every agent page.

    planned
  4. W1·23
    Live generator and reactor outage feed

    An outage at a research reactor moves clinics worldwide within days. A dated feed of announced outages and campaign schedules would be read every week.

    proposed
  5. W1·19
    Isotope supply page with the real constraints

    Reactor campaigns, cyclotron capacity, generator parents, enrichment and 68Ge or 99Mo shortages are the field's binding constraint. One page that says who makes what and what is tight.

    shipped

Practice 4

  1. W2·27
    Centre directory for theranostics

    Which centres run radioligand therapy and which trials they host, so a clinician or patient can find the nearest one.

    Editorial status was “shipped”, but institutions with people is 19/60 (31.7%, target 80%). Claim it.

    needs work
  2. W2·24
    Guideline and appropriate-use index

    SNMMI, EANM and specialty-society guidance per agent and per indication, linked from the agent page with the year each was issued.

    building
  3. W2·25
    Radiation-protection and release-criteria reference

    The practical questions after a therapy — waste, family contact, travel, pregnancy, breastfeeding — differ by nuclide and jurisdiction and are badly documented online.

    planned
  4. W2·26
    Authorised-user and licensing map

    Who may hold and administer therapeutic activity differs by country and by state; it is the quiet limit on how fast therapy scales.

    proposed

Pipeline 6

  1. W2·32
    Conference digests (SNMMI, EANM, ASCO GU, ESMO)

    Within a week of each meeting, update the affected records and publish the diff.

    building
  2. W2·28
    Trial tracker with structured outcomes

    Every trial record carries its registry id, phase, design and the outcome numbers as published, not as press-released.

    Verified: trials with structured outcomes is 38/39 (97.4%, target 95%).

    shipped
  3. W2·29
    Automated ClinicalTrials.gov ingestion

    A scheduled job pulls phase 2/3 trials for every agent and target in the corpus and flags new ones for review.

    Verified: products with a clinicaltrials.gov snapshot is 100/100 (100%, target 95%).

    shipped
  4. W2·30
    Readout and regulatory calendar

    Expected readouts, PDUFA dates, CHMP opinions and advisory committees for radiopharmaceuticals on one timeline.

    shipped
  5. W2·31
    Failure museum

    Programmes that stopped — and why — are the most useful and least published data in the field.

    shipped
  6. W2·33
    Key-paper library with citation counts

    The twenty papers that define each sub-field, with DOIs, so a newcomer can read the actual literature rather than a summary of it.

    Verified: key papers with a citation count is 23/23 (100%, target 95%).

    shipped

Openness 4

  1. W2·34
    Everything under version control, corrections welcome

    Records are TypeScript files in a public repo: anyone can open an issue or a pull request against a specific claim.

    Editorial status was “shipped”, but records with git provenance is 67/862 (7.8%, target 95%). Claim it.

    needs work
  2. W2·37
    Published gaps list

    Say out loud what is missing — kinds with too few records, records with no source, isotopes with no agent — instead of hiding it.

    Editorial status was “shipped”, but records with three or more relations is 770/850 (90.6%, target 95%). Claim it.

    needs work
  3. W2·35
    Named expert review per page

    A radiochemist, a nuclear-medicine physician and a medical physicist signing off the pages in their area, with the date and their conflicts declared.

    Verified: records with a review badge is 0/850 (0%, target 10%).

    planned
  4. W2·36
    MCP server and CLI

    The corpus should be usable by an AI assistant or a terminal without scraping the site.

    shipped

Patients 3

  1. W3·38
    Patient-facing explainers per therapy

    What the treatment day looks like, how many cycles, what the scan shows, what the side effects actually are — in plain language, sourced to the label.

    planned
  2. W3·39
    Translated TL;DRs

    Nuclear medicine is global and supply is not. Translate the one-sentence layer first, where the value is highest and the cost lowest.

    proposed
  3. W3·40
    Cost and access notes where a source states them

    List price, reimbursement status and per-cycle cost decide who gets treated; record them only where a public source says so.

    proposed

Chemistry 2

  1. W3·41
    Chelator and linker chemistry reference

    DOTA, DOTAGA, macropa, sarcophagine, HYNIC: which nuclide each holds, at what temperature, under what patent.

    Editorial status was “building”, but products with a molecule or an explained placeholder is 64/100 (64%, target 95%). Claim it.

    needs work
  2. W3·42
    Target prevalence with sources

    Expression frequency per indication decides who is eligible; each number needs a citation, not a round figure.

    Editorial status was “building”, but targets with sourced prevalence is 11/36 (30.6%, target 95%). Claim it.

    needs work

Community 2

  1. W3·44
    Who's who of the field

    The people whose papers and programmes moved the field, linked to their institutions and their key papers.

    Verified: people with papers is 45/47 (95.7%, target 95%).

    shipped
  2. W3·45
    Newsletter and dated digests

    A short, dated summary of what changed in the corpus, so a reader can follow the field by following the diff.

    shipped

C. Method

Two loops. Expansion adds records and features in waves and gates them; currency refreshes what the world has changed and audits what we already say. Each card says what runs, how often, where the code is, and what would make it better.

How expansion works

Gap audit against the goal

Enumerate the universe (isotopes in clinical or trial use, FDA/EMA approvals, indications, guideline tables), diff it against the corpus, and write the gaps down with size and owner. The gauges above are the automated half; an editorial pass at the start of each wave is the other.

How often
Gauges: every build. Editorial audit: at the start of each wave.
Where
src/lib/health.ts/gaps/
Better if
Open one GitHub issue per open gap automatically, so the audit becomes a work queue rather than a document.

Per-kind agents in parallel with disjoint file ownership

A wave is split into agents, each owning a set of data files nobody else touches (cancer deep dives, gap-fill, idea waves, people, key papers). They run in separate git worktrees, so merges are mechanical: no two agents edit the same file.

How often
Per wave; seven agents in the last one.
Where
src/data/spikes/src/data/gap-fill.tssrc/data/ideas-waves/
Better if
An ownership manifest checked in CI, so a change to a file outside an agent's allotment fails the build instead of a merge.

Validate, typecheck, lint, test, build gates

Every push and pull request runs the schema validator (ids, dangling references, duplicates), the TypeScript compiler in strict mode, ESLint, the corpus tests (every TL;DR readable, every reference resolving), and a full static build.

How often
Every push to main and every pull request.
Where
.github/workflows/ci.ymlscripts/validate.tssrc/lib/graph.test.ts
Better if
Turn each gauge target into a ratchet: a pull request may not lower a metric that is already above its target.

Suggest-an-edit issues

Every object page has a form that opens a prefilled GitHub issue (entity, field, proposed value, source, who you are, conflicts) or a direct link to the record's line for a pull request. Organisations may edit their own records under their name.

How often
Continuous; triage within a week.
Where
SuggestEdit.tsxissue templates/suggest/
Better if
A bot that converts an accepted issue into a pull request against the exact line, with the source appended to `links`.

Review tracks

Clinical, scientific, regulatory, patient-advocate, and organisation self-edit tracks, each with named reviewers and a conflict-of-interest field shown next to the badge. The badge component is built; the reviewer list is not yet populated.

How often
On merge of any factual change; badges dated.
Where
.github/REVIEWERS.mdsrc/data/reviews.tsReviewBadge.tsx
Better if
Recruit one named reviewer per cancer and per front; the “Records with a review badge” gauge above is at zero until then.

How currency works

Trials from ClinicalTrials.gov

For every product, pull phase 2/3 studies from the ClinicalTrials.gov v2 API into public/trials/, then open a pull request with the diff for review.

How often
Weekly (Mondays 06:17 UTC) via GitHub Actions; also on demand.
Where
refresh-trials.ymlscripts/fetch-trials.ts
Better if
Cover every product, not just the first batch (see the trials-snapshot gauge), and flag new recruiting trials as candidate trial records.

Literature from Europe PMC

Paper counts per year and the latest titles for every drug, target, cancer, and technology into public/papers/; the /papers/ page recomputes the fastest-growing topics from the snapshot.

How often
Weekly (Tuesdays 05:41 UTC) via GitHub Actions.
Where
refresh-papers.ymlscripts/fetch-papers.ts
Better if
Add the records created since the last run automatically (the papers-snapshot gauge shows the backlog) and tune query terms for ambiguous names.

Fact check against openFDA and ClinicalTrials.gov

Products recorded as US-approved are checked against openFDA labels; trials with an NCT id against the registry's overall status. Mismatches land in public/factcheck.json and on /audit/; confirmed errors go to the corrections log.

How often
Weekly (Mondays 06:41 UTC), in the same job as the audit and provenance refresh.
Where
factcheck.ymlscripts/factcheck.ts
Better if
Add EMA, MHRA, PMDA, and NMPA checks so the regional approvals table is verified, not just the US column.

Audit for staleness and contradictions

A pure function of the corpus: status fields that disagree with approvals, positive trials without results, standard-of-care rows citing withdrawn items, future years, duplicate names, and every record's asOf age.

How often
Every build (rendered at /audit/) and weekly in the fact-check job.
Where
scripts/audit.ts/audit/
Better if
Treat high-severity findings as build failures once the backlog is cleared.

Provenance from git

git blame over every data file maps each record to the last commit that touched it, so every page can say who edited it, when, and link the diff.

How often
Weekly in the fact-check job; needs full history, so it does not run at build time.
Where
scripts/provenance.tsProvenanceLine.tsx
Better if
Run it in the deploy step so new records never wait a week for a provenance line (see the provenance gauge).

Molecule structures from PubChem and PDB

Resolves each product's structure definition to a compact JSON under public/structures/; small molecules rotate on their pages, biologics get an explained placeholder.

How often
Manual: `npm run fetch:structures`. No scheduled workflow yet.
Where
scripts/fetch-structures.tssrc/data/structures.ts
Better if
Schedule it, and derive the PubChem lookup from the product name so new small molecules do not need a hand-written entry.

Logos from Wikidata

For every company, institution, and collection: Wikidata match with domain verification, then Commons logo, with favicon fallback, into public/logos/.

How often
Manual: `npm run fetch:logos`. No scheduled workflow yet.
Where
scripts/fetch-logos.ts
Better if
Schedule it monthly and fail loudly when a logo's licence is missing.

Research output from OpenAlex

Oncology works and citations per institution and per country (OpenAlex subfield 2730), feeding the university and country rankings, with the query published.

How often
Manual: `npm run fetch:openalex` and `npm run fetch:countries`. No scheduled workflow yet.
Where
scripts/fetch-openalex.tsscripts/fetch-countries.ts
Better if
Schedule quarterly; store the run date on the ranking pages so readers can see how fresh the counts are.

asOf on every record

The schema requires an asOf date on every record: when its facts were last checked. The audit lists the oldest; the staleness gauge above counts anything older than 60 days.

How often
Set by whoever edits the record; checked every build.
Where
src/lib/schema.ts
Better if
A monthly “re-verify” queue of the 50 oldest records, assigned by kind.

Corrections log and changelog

Every confirmed factual error is logged with what was wrong, how it was found, and the fixing commit; every release is summarised in the changelog, which doubles as the weekly newsletter.

How often
Corrections as they are confirmed; changelog per release.
Where
CORRECTIONS.mdCHANGELOG.md
Better if
Generate the changelog's data section from the git diff of src/data/ so no change goes unannounced.

D. Propose an idea

Ideas live in src/data/hub-ideas.ts as records with a title, a one-line why, a theme, a status, and optionally the id of a gauge that measures whether the claim holds. To add one:

  1. Open a roadmap issue with the idea, why it matters, and how we would know it is done. Argue for it there.
  2. Or add the record directly in a pull request. If the idea makes a measurable claim, add a metric for it in src/lib/health.ts and reference it with metric, so the roadmap can never call it shipped when it is not.
  3. Statuses are editorial; the gauges are not. A shipped idea with a failing gauge shows as needing work until the corpus catches up.