Searches run (live PubMed, esearch/esummary/efetch + MCP PubMed tools): staging statement (Insel 2015); teplizumab TN-10 and PROTECT; closed-loop pivotal trial and AID meta-analyses; global T1D burden models (Gregory 2022, GBD 2021); Ziegler 2013 autoantibody progression; DCCT 1993 and DCCT/EDIC 30-year follow-up; Rawshani mortality series; Livingstone life expectancy; T1D Exchange registry; time-in-range consensus; zimislecel stem-cell islets; verapamil trials (adult + CLVer); tight-control CLVer arm; Fr1da screening; pre-stage-3 monitoring consensus; DKA-at-diagnosis meta-analysis; impaired hypoglycemia awareness; diabetes distress and fear-of-hypoglycemia reviews; ADA Standards of Care 2026 sections.
Verification: all cited PMIDs retrieved from live PubMed during this session; abstracts fetched for all quantitative claims. Volume/page details included only where returned by the live query.
Next: full build queued per CONDITIONS-ROADMAP.md — build remaining 16 wiki pages, literature layer (BIBLIOGRAPHY, guidelines REGISTRY, STATISTICS, patient-voice), and verify the trials landscape against ClinicalTrials.gov.
Built: all 17 canonical wiki pages; preserved and extended the seeded overview and four pre-existing draft pages; all pages remain draft pending a separate audit session.
Literature layer: 96-record deduplicated bibliography; five landmark notes (DCCT, Ziegler seroconversion, teplizumab, iDCL closed loop, zimislecel); 15-record guideline registry; 31-row statistics reference; patient-voice method, themes, sources and four live-verified organizations.
Live PubMed searches: staging/natural history; global epidemiology, mortality and DKA; human pancreatic pathology, insulitis and β-cell stress; HLA/GWAS/genetic risk scores; enterovirus, vitamin D, diet and microbiome; Fr1da/monitoring; DIAMOND, IMPACT, iDCL, pediatric closed loop and bionic pancreas; teplizumab, rituximab, abatacept, ATG, oral insulin and verapamil; donor-islet and stem-cell replacement; impaired awareness, HARPdoc and glucagon; DCCT/EDIC; distress, eating, transition, stigma and technology disparities; biomarker and guideline topics. All final page PMIDs were re-resolved in a batch E-utilities verification pass.
ClinicalTrials.gov v2 searches: T1D condition plus targeted teplizumab, abatacept, rituximab, ATG, oral insulin, zimislecel/VX-880, VX-264, bionic-pancreas, closed-loop and HARPdoc queries. Fifteen returned NCT records were retained across the condition corpus.
Public websites fetched: NICE NG17 and NG18; Diabetes UK; T1International; American Diabetes Association; Diabetes Canada. Breakthrough T1D/JDRF, Life for a Child and several regional sites returned access-control errors and were omitted from the verified organization table.
Open questions: replaced seed list with OQ-1 through OQ-20 in two tiers and a ten-row “Dots not yet connected” table.
Follow-up: run the required fresh-session audit; verify every quantitative claim against abstracts/full text, check all NCT statuses again, expand guideline and patient-organization geographic coverage, and promote only clean pages to curated.
Known limitations: page lengths are dense but mostly below the aspirational 150-line target; the bibliography has 95 unique papers rather than the reference condition’s ~350; organization coverage is English-language and high-income weighted. No inaccessible organization or unreturned identifier was treated as verified.
Deepened all 17 canonical wiki pages in place: retained the verified 2026-08-28 text and added quantitative natural-history, incidence, mechanism, screening, biomarker, technology, hypoglycemia, complication, immunotherapy, replacement, guideline, psychosocial, patient-experience, and acute-safety evidence. Final page lengths are 150–210 lines; every page contains at least 25 unique PMID citations (range 25–41).
Evidence emphasis: added landmark and confirmatory cohorts, effect sizes with confidence intervals, assay and threshold limitations, adult-versus-child calibration, contemporary comparator problems, intervention nonresponse, harms, implementation inequity, patient-reported outcomes, and explicit tables preserving conflicting interpretations.
Live PubMed verification: ran topic searches and record retrieval through PubMed E-utilities on 2026-08-30. The canonical wiki set now contains 256 unique PMIDs. A reconciliation check found zero wiki PMIDs absent from literature/BIBLIOGRAPHY.md; every newly added record has a cited-by mapping.
Literature layers: expanded the statistics file from 31 to 72 table rows; expanded the guideline registry from 15 to 22 records with topic supersession and scope comparison; sharpened OPEN-QUESTIONS.md from 20 to 26 stable questions and from 10 to 15 “dots not yet connected.”
Curation state: retained draft status on every wiki page. This was a deepening/build session, not the independent fresh-session audit required for promotion to curated.
Remaining limitations: adult prospective staging, general-population screening effectiveness, rare-event safety under modern AID, long-term disease-modification durability, representative patient-experience sampling, and contemporary controlled comparisons for cell replacement remain evidence gaps rather than missing prose.
Scope checked: all 17 canonical wiki pages, OPEN-QUESTIONS.md, all 12 Markdown artifacts in literature/, the condition index, and the roadmap entry. The audit inspected 1,580 PMID mentions (1,087 in wiki pages, 416 in literature artifacts, 77 in OPEN-QUESTIONS.md), representing 259 unique PubMed records; 49 NCT mentions representing 16 unique ClinicalTrials.gov records; and eight public URLs (six organizations plus NICE NG17/NG18).
Live verification: all 259 current PMIDs resolved in a fresh PubMed E-utilities efetch; 187 linked PMC full texts were additionally retrieved for quantitative cross-checking. All 16 current NCT IDs resolved through exact ClinicalTrials.gov v2 study endpoints. All eight public URLs returned HTTP 200 on recheck. No identifier remained unresolved.
Citation errors found and fixed: 32 discrete citation-layer defects — one wrong claim-to-PMID pairing (the ISPAD diabetes-education row incorrectly cited the hypoglycemia chapter), 13 distinct misattributed or bibliographically incorrect records propagated across inline prose/reference lists/bibliography, and 18 stale cited by mappings in BIBLIOGRAPHY.md. Corrections included first authors, publication years, journals, article numbers/pages, and cited-by destinations; the bibliography now reconciles exactly to all 259 records present in the wiki set.
Claim corrections and overstatement control: corrected the education evidence row; changed several patient-experience attributions to the actual first authors returned by PubMed; narrowed “technology access” to the insulin-and-device access actually studied; and retained uncertainty around uncontrolled, post hoc, rare-event, and selected-cohort findings. Quantitative claims were checked against abstracts and, where available, PMC full text; no unresolved contradictory number remained.
Absence/evidence-gap searches rerun: (1) durable immune tolerance — a newly retrieved six-person “immune education” pilot showed 91–100% baseline C-peptide AUC at 12 months but 44–56% at five years, so the dated gap remains; (2) immune-isolated/hypoimmune replacement — NCT06239636 was found recruiting, establishing human testing but not durable efficacy; (3) vitamin D prevention — a 2025 meta-analysis found no significant reduction in islet autoimmunity or T1D; (4) 2024–2026 ISPAD topic replacements — title/date searches found no replacement for the 2022 insulin, nutrition, hypoglycemia, fasting, or education chapters as of 2026-08-30; (5) previously inaccessible organization sites — Breakthrough T1D and Life for a Child now resolved and were added.
Result: all 17 pages meet the required structure and 150–400-line range (150–212), contain 25–41 unique verified PMIDs, have complete body-to-reference coverage, resolve all relative wiki links, and contain zero [unverified] flags. All 17 were promoted to status: curated; INDEX.md now reports audited status and current counts; CONDITIONS-ROADMAP.md was advanced to audited.
Remaining flagged: none. Remaining research limitations are stated positively as dated evidence gaps rather than unresolved verification flags. Future audits should recheck fast-moving trial statuses (especially NCT06239636, whose recruiting label was last sponsor-verified in 2024-12), annual ADA/ISPAD supersession, and durability/safety updates for immune and cell-replacement interventions.