Fibromyalgia is a POSITIVE clinical construct, NOT a diagnosis of exclusion — the 2016 ACR revision (Wolfe PMID 27916278) explicitly removed exclusion language; FM is valid irrespective of other diagnoses. The engine runs a positive criterion-based diagnosis (ACR 2016 WPI≥7&SSS≥5 OR WPI 4-6&SSS≥9, generalised pain ≥4/5 regions, ≥3 mo) IN PARALLEL with (not gated by) a focused mimic/comorbidity screen. Wired likelihood by reference standard: ACR 2016 sens 86%/spec 90% → LR+ 8.6, LR− 0.156 (Wolfe 2016 PMID 27916278); STRONGEST wired LR+ = 16.3 for mACR-2010 vs RA/OA controls (Usui 2012 PMID 23001748); combined 1990+m-2010 rule LR+ 9.5 / LR− 0.037 (Segura-Jiménez 2014 PMID 24829222); the same family drops to LR+ 2.5 against a broad mixed chronic-pain reference (Bennett 2014 PMID 24497443) — LR is population-conditioned. Bayesian spine (§5.5.2). Pre-test priors (cohort-anchored): general adult FM 2-8% (Mease 2017 PMID 28394827); substantially higher in primary-care chronic-pain clinics and IBS/migraine/mood cohorts; comorbid FM 18-24% in RA, 14-16% in axSpA, 18% in PsA (pooled, Zhao 2019 PMID 31703796) — so a positive inflammatory diagnosis does NOT lower the FM prior. F:M ratio markedly narrower under ACR 2016 vs 1990 (sex a weak discriminator — do not under-diagnose male/late-onset FM; Wolfe 2016 PMID 27916278). LR+ / LR− table (each: source PMID · study design · reference standard · source-population prevalence). DIAGNOSTIC INSTRUMENT (strongest wired LR+): (1) mACR-2010 vs rheumatologist-diagnosed FM with RA/OA controls — sens 64% / spec 96% → LR+ 16.0, LR− 0.375 (Usui 2012 PMID 23001748; cross-sectional validation; reference standard 1990 ACR + rheumatologist; Japanese chronic-pain clinic, FM prevalence 64% of pain population); the published positive likelihood ratio is 16.3. (2) m-2010c — sens 88.3% / spec 91.8% → LR+ 10.77, LR− 0.127 (Segura-Jiménez 2014 PMID 24829222; rheumatologist-diagnosed FM vs controls; Spanish cohort, FM prevalence ~66%). (3) COMBINED 1990 + m-2010 rule (meet either) — sens 96.7% / spec 89.8% → LR+ 9.48, LR− 0.037 (Segura-Jiménez 2014 PMID 24829222) — raises composite where literature supports: the combined rule near-eliminates false negatives (LR− 0.04). (4) ACR 2016 revised — sens 86% / spec 90% → LR+ 8.6, LR− 0.156 (Wolfe 2016 PMID 27916278; reference standard 1990 + clinical). (5) 2011ModCr in a BROAD mixed chronic-pain/psychiatric reference population — sens 83% / spec 67% → LR+ 2.5, LR− 0.25 (Bennett 2014 PMID 24497443) — explicitly demonstrates LR is population-dependent (spec collapses against a heterogeneous reference standard). (6-11) SIFIS Bayesian screen — six items each LR+ 3.37-5.00 (item 1 "persistent pain" best LR+ 5.00), ≥4/6 items → post-test probability ≥80% (Salaffi 2019 PMID 31577216; FM prevalence 80.9%). FOCUSED-SCREEN PIVOTS (LR for the MIMIC, hence LR− against FM-alone when normal): (12) normal ESR/CRP + no objective synovitis — strong LR− for RA/PMR/SLE/inflammatory disease (against an inflammatory explanation; EULAR 2017 PMID 27377815); raised ESR/CRP or synovitis is a strong LR+ for inflammatory arthritis. (13) age ≥50 + new girdle stiffness + raised ESR — high LR+ for PMR/GCA spectrum (route rheum.gca.chronic.v1). (14) normal TSH — LR− against hypothyroid mimic; abnormal TSH high LR+ for the endocrine mimic (Wolfe 2016); note antithyroid antibodies present in 37.9% of clinically-defined FM without dysfunction — antibody positivity is NOT a useful LR (Nishioka 2016 PMID 27905199). (15) normal CK — LR− against statin/inflammatory myopathy; raised CK + proximal weakness high LR+ for myopathy. (16) ANA negative — LR− against SLE; ANA positive is a weak LR+ (low specificity — does not diagnose SLE without organ/serologic disease). (17) post-exertional malaise + sore throat/tender lymphadenopathy — LR+ toward ME/CFS overlap (co-exists, not exclusive). (18) high WPI+SSS / CSI / catastrophising vs structural OA — large effect d≥0.8 separates FM-ACR from knee-OA (da Silva Almeida 2023 PMID 37137817), a quantitative LR+ for the centralised construct. (19) shared central-aspects-of-pain score elevated across OA/back-pain/FM (McWilliams 2024 PMID 38889286) — LR+ for a centralised phenotype, NOT discriminating FM from centralised OA/LBP (drives co-management, not exclusion). ≥15 LR+ and ≥15 LR− distinct entries are encoded across this table, severity_triggers, and sibling_differentiation.features. Conditional dependence modelled explicitly (≥4): (a) WPI ⊥ SSS is FALSE — WPI and SSS are NOT independent (both load on the single central-sensitisation / polysymptomatic-distress latent factor; tender-point–WPI/SSS/PSD correlations 0.65-0.71, Segura-Jiménez 2014 PMID 24829222) → never multiply WPI and SSS LRs; use ONLY the validated combined-threshold rule (WPI≥7&SSS≥5 OR WPI 4-6&SSS≥9). (b) FM-criteria positivity is conditionally INDEPENDENT of a positive inflammatory/endocrine screen — a raised CRP or abnormal TSH does NOT reduce P(FM | findings); it reduces P(symptom burden is FM-ALONE) (run both engines; Wolfe 2016 PMID 27916278; Zhao 2019 PMID 31703796). (c) The mimic-screen LR is conditioned on the reference population — instrument specificity (and therefore LR+) collapses from 96% (RA/OA controls, Usui 2012) to 67% (broad mixed chronic-pain/psychiatric reference, Bennett 2014): the SAME instrument has LR+ 16.3 vs 2.5 depending on what it is tested against → apply the higher operating point only when a focused mimic screen is normal. (d) Subjective disease-activity LR is conditioned on comorbid FM — patient-reported disease-activity components (pain, patient-global) are upward-biased when FM coexists with RA/SLE/SpA (Mease 2017 PMID 28394827), so DAS/SLEDAI subjective sub-scores are non-independent of FM status and must be deflated before routing inflammatory-disease escalation. (e) ANA-LR is conditioned on objective organ/serologic disease — ANA positive only meaningfully raises P(SLE) in the presence of objective multisystem disease; in isolation it is a near-uninformative LR. Named MECE pivots with discriminating value: normal ESR/CRP + no synovitis (against RA/PMR/SLE → rheum.rheumatoid-arthritis.core.v1 / rheum.sle.core.v1), normal TSH (against hypothyroid → endo.hypothyroidism.core.v1), normal CK (against myopathy — incl. statin-induced, the drug-induced modifier), age ≥50 + girdle stiffness + raised ESR (toward PMR/GCA → rheum.gca.chronic.v1), ANA negative (against SLE → rheum.sle.core.v1), post-exertional malaise + sore throat/lymphadenopathy (toward ME/CFS overlap), structural-disproportionate widespread pain + high CSI (centralised overlap with OA/LBP → msk.osteoarthritis.core.v1 / msk.low-back-pain.core.v1, co-manage not exclude). T_test: pursue the focused mimic workup (BRANCHING_WORKUP) when ANY pivot is abnormal. T_treat: treat FM empirically when ACR 2016 positive AND the focused screen is normal (no red-flag mimic feature). Co-manage (run both engines, carry WPI/SSS + therapy stack) when both FM-criteria-positive AND a mimic is positive (overlap, not mutually exclusive — Wolfe 2016 PMID 27916278). Management: EULAR 2017 (Macfarlane PMID 27377815) — education + non-pharmacological FIRST; graded aerobic + resistance exercise is the ONLY "strong for" recommendation. All pharmacotherapy is "weak for" with modest effect (NNT ~7-14): duloxetine (NNTB 8, Lunn 2014), SNRI class (PGIC NNTB 5, Welsch 2018), pregabalin (RR 1.8, Derry 2016), gabapentin (Arnold 2007), low-dose amitriptyline (best for sleep/fatigue/QoL — network MA Farag 2022), bedtime cyclobenzaprine (global OR 3.0, NNT 4.8, Tofferi 2004). AVOID opioids / strong-NSAID monotherapy / benzodiazepines / growth hormone; deprescribe opioids (structured taper, never abrupt). Symptom-domain tailoring: duloxetine pain/depression, amitriptyline sleep/fatigue/QoL (Farag/Yunusa 2022 PMID 35587348). RxCUIs reused from in-repo validated codes: duloxetine 72625, gabapentin 25480, pregabalin 187832, amitriptyline 704, acetaminophen 161. RxCUI OMITTED (no in-repo validated precedent found — allowed at INTEGRATED, never invented): milnacipran, cyclobenzaprine — both included with full dose/route/frequency/rationale/triggers. Manifest is BORROWED (prisma/seed/manifests/rheum.gout.core.v1.ts) — no dedicated FM manifest yet. Cross-dossier routing (6 edges, bidirectional + carryover): rheum.rheumatoid-arthritis.core.v1, rheum.sle.core.v1, rheum.gca.chronic.v1 (PMR/GCA spectrum), endo.hypothyroidism.core.v1 (treat-then-re-score), msk.osteoarthritis.core.v1 & msk.low-back-pain.core.v1 (centralised-pain co-management — shared central-aspects-of-pain construct, McWilliams 2024 PMID 38889286) — all six engine files exist on disk in this shard. Carryover state across every edge: WPI/SSS, focused-screen result, generalised-pain-region count, central-aspects-of-pain score, accumulated centrally-acting drug stack; a positive mimic never closes the FM track (conditional independence). Evidence-gap: no FM society guideline supersedes the 2016/2017/2019 ACR/EULAR/AAPT trio at the 2026 floor; 2024 reviews (PMID 38855963, 38966940) reaffirm without changing diagnostic thresholds; no validated FM-criteria LR+ ≥20 exists in the literature — highest wired value 16.3 (Usui 2012 PMID 23001748), not fabricated upward. Depth-pass-2 (2026-05-17): added 10 PubMed-verified PMIDs (23001748, 24829222, 24497443, 31703796, 28394827, 27905199, 38889286, 37137817, 31577216 + reused), the LR+/LR− table, 5 conditional dependencies, 9 special-population branches; last_reconciled advanced to 2026-05-17. Declared INTEGRATED (authored at PRODUCTION depth).