[2026-10-01] SELECTION RECORD — DOI resolution for three tokenized gaps (not a discovery sweep)

(1) Query. Two openalex title searches (auto-logged above) to resolve DOIs for placeholder registry rows: Gregg 2016 Look AHEAD weight-loss-magnitude CV outcomes; IARC 2016 coffee/mate/hot-beverages re-evaluation. Hidayat 2022 red-meat-substitution DOI already on file (10.1093/nutrit/nuac017).

(2) Results. Gregg 10.1016/s2213-8587(16)30162-0 (oa=True). IARC-summary 10.1016/s1470-2045(16)30239-x (Lancet Oncol, oa=False; full IARC Monograph Vol 116 free at publications.iarc.fr). No zero-hit queries.

(3) Adjudication. Not a screening pass — these three are already-identified deliverable gaps being TOKENIZED as [AWAITS (slug)] holds (Body Fat / Coffee / Meat), not newly selected. DOI resolution only, to mint accurate placeholder rows. Able/want read recorded in acquisition-worklist.md 2026-10-01.

(4) Acquisition-requested. 0 fetched this pass (tokenization, not acquisition). Forward plan: Gregg = acquire (OA, one fetch); Hidayat = maintainer-drop; IARC coffee = skip unless a cancer-axis question later demands it (type-E echo of held Poole). No downgrade, no substitution.

[2026-10-04] SELECTION RECORD — discover: exercise programming for the obese/untrained/sedentary/middle-aged stratum

Context. Framing + source-discovery for a down-the-line deliverable: minimum effective / minimum maintenance exercise programming for obese, sedentary, middle-aged individuals (CRF, functional fitness, QoL, independence — NOT high-performance). ~29 openalex gold-tier queries (logged above). Gap cells probed: HIIT-vs-MICT in obesity · aerobic/VO2max dose · resistance-training dose · running / overuse injury risk · tendon / connective-tissue adaptation to loading · detraining / maintenance-dose shape · walking -> function · sedentary-behaviour · exercise prescription in obesity.

(3) Adjudication + reason (per candidate).

SELECTED — Tier 1 (genuine 0-coverage gap, highest ratchet):

  • Böhm/Wiesinger - Human tendon adaptation to mechanical loading 2015 (10.1186/s40798-015-0009-9, gold MA, 465 cites). NOVEL (discover —novelty confirmed; no held tendon/connective-tissue page). Fills the Q3 cell nothing covers: how, and how much, tendon adapts to load before it tolerates impact — the deconditioned-joint / progressive-overload question the user flagged. The ONLY source on tissue-adaptation dose. -> FETCHED to inbox/.
  • Hespanhol/Sports Med - Running-related injury risk factors SR 2019 (10.1007/s40279-019-01110-z, SR, 264 cites). Fills the “what predicts running injury” leg of Q3; population mostly established runners, so transfer-to-obese-novice stays a NAMED GAP on Exercise Interventions and Sports Injury Prevention — but it is the best available. -> maintainer-drop (Springer wall).
  • sms - Residual effects of exercise after training cessation, older adults SR 2025 (10.1111/sms.70010, SR). Directly the maintenance-dose / detraining SHAPE cell (how fast is CRF/ function lost without the stimulus — the “maintenance over a long horizon” question). -> maintainer-drop (Wiley block).

SELECTED — Tier 2 (obese-stratum programming synthesis; likely type-F refinement of the held CRF page):

  • Wewege - Exercise prescription for body composition + CRF in overweight/obese adults 2020 (10.1111/obr.13137, 195 cites). A “what to program for the obese” synthesis = the deliverable’s nucleus; novelty-read vs Measuring and Raising Cardiorespiratory Fitness owed at ingest. -> maintainer-drop (Wiley block).
  • Jiménez-Pavón / MSSE - Moderators of resistance-training effects in overweight/obese adults 2022 (10.1249/mss.0000000000002984, MA). Route-(b) effect-modification in the exact stratum. -> maintainer-drop (paywall).

DEFERRED (candidate, lower marginal rank — acquire only if the cell stays thin):

  • Walking-only -> physical function / falls, older 2024 (10.1123/japa.2023-0183, paywalled) and Walking + health outcomes, older adults 2026 (10.1177/08901171261438402, OA). The Q1 reframe toward function/QoL/independence promotes walking->FUNCTION (held page covers only walking->mortality). DEFER: low cites, young literature; revisit when the function-outcome cell is scoped.
  • “Use It or Lose It” — resistance-training cessation MA 2022 (10.3390/ijerph192114048, OA). RT- specific detraining; secondary to the broader sms 2025 SR. DEFER.
  • Combined aerobic+resistance -> glycemic/body-comp MAs (10.1016/j.ypmed.2016.10.020; 10.7717/peerj.17525). Metabolic-outcome cell, overlaps held T2D/metabolic pages; not this deliverable’s aerobic-CRF core. DEFER.
  • Progressive resistance training -> disability/sarcopenia, older 2022 (10.1007/s40279-022-01675-2; 10.1016/j.exger.2022.111767). The “progressive overload” principle lives here but likely overlaps held grip-strength/muscle-mass/RT->function pages; novelty-check vs held RT pages before acquiring. DEFER.

REJECTED:

  • HIIT-vs-MICT in obesity cluster — Wewege 2017 (10.1111/obr.12532, 670 cites) + 2016/2020/2021/ 2026 MAs. SATURATED: the held Poon 2024 umbrella on Measuring and Raising Cardiorespiratory Fitness already pools HIIT>=MICT (modest, adherence-bound). Expectancy test fails — a 2017 MA will not move a recommendation anchored on a 2024 umbrella. The saturation IS the finding: the HIIT-vs-walking cell is covered; do not re-acquire.
  • Concurrent (strength+endurance) training MAs (10.1080/02640414.2017.1364405 etc.) — interference/performance concern, frontier for the trained stratum, off the obese-sedentary MED scope.
  • Sedentary-behaviour digital/wearable interventions (10.2196/14343 etc.) — adherence channel, tangential to dose.
  • “Minimum exercise dose” for bone remodeling (10.7759/cureus.25993) — osteoporosis-specific, off-scope.

ATTRIBUTED ZEROS (0-hit = query artifact, not literature absence):

  • “zone 2 training” -> 1 surgery false-positive. “Zone 2” is colloquial; the SR/MA literature titles it “moderate-intensity” / lactate-threshold / aerobic-base / maximal-fat-oxidation. CORRECTION (2026-10-04): an earlier draft of this line claimed the content is “already held in the CRF intensity-zone table” — that was an OVERCLAIM. Verified by grep: the held CRF page holds only the HIIT-vs-MICT story + %HRR zones; it does NOT hold the two-regime physiology (sub-LT1 oxidative / fat-ox / mitochondrial / type-I vs at/above-LT2 glycolytic / type-II). The lactate-threshold / fat-oxidation / polarized model is a GENUINE GAP. Acquisition cell added to worklist (FatMax-training source; polarized marked frontier). Not literature absence — a real fabric hole.
  • “non-exercise activity thermogenesis SR”, “exercise maintenance dose SR” -> 0. NEAT/maintenance-dose are largely mechanistic/observational (Levine lineage), rarely SR/MA-shaped; plausible genuine SR/MA thinness. Named GAP; retry type-unfiltered if the NEAT lever is pursued.
  • “novice runner injury”, “exercise VO2max obese”, “exercise volume frequency cardiorespiratory”, “detraining cardiorespiratory MA” -> 0. Over-constrained multi-term/singular title.search; the cells are covered by the broader running-injury / training-cessation / dose-response queries that DID hit.

(4) Acquisition-requested. 1 fetched -> inbox/ (Böhm tendon 2015, OA). 4 -> maintainer-drop, no downgrade (Springer/Wiley/paywall): Hespanhol running-injury 2019, sms training-cessation 2025, Wewege obesity-prescription 2020, MSSE RT-moderators 2022. Handoff: inbox PDF -> process inbox -> delegated ingest-seq (never main-agent); maintainer-drops await the drop. Deferred candidates stay worklist rows, not fetched.

[2026-10-04] addendum — modality + injury/progression axes (same discover pass)

Two further axes the maintainer asked for: (a) modality (calisthenics/cycling/swimming/rowing/walking/ running/RT) x adherence / injury / CRF / RPE; (b) tissue-specific running injury + progression rate (the “couch-to-10k, obese running too early risks ligament/metatarsal/Achilles injury” belief).

(3) Adjudication.

SELECTED (new cells, folded to worklist 2026-10-04 block):

  • Training-load changes -> running injury SR 2018 (10.26603/ijspt20180931) — THE progression-rate (“too much too soon”) evidence; the couch-to-10k belief’s core. HIGH.
  • Lower-extremity stress-fracture risk factors in runners MA 2015 (10.1136/bjsports-2015-094828) + metatarsal stress-fracture SR 2024 (10.3389/fbioe.2024.1435807) — the bone-stress/metatarsal arm.
  • Running-related injury risk factors SR 2019 (10.1007/s40279-019-01110-z) — general RRI predictors.
  • BMI -> running injury SR 2025 (10.21203/rs.3.rs-6311307/v1) — on-stratum but researchsquare PREPRINT, tier-caution, verify the peer-reviewed version before ingest.
  • Exercise TYPE & duration -> CRF MA 2023 (10.3389/fcvm.2023.1190187) — the modality-vs-CRF cell.
  • Exercise referral schemes effect MA 2011 (10.1136/bmj.d6462) — real-world adherence in the sedentary/referred stratum (Layer-3 anchor).
  • Affective responses to HIIT MA 2020 (10.1080/17437199.2020.1728564) — intensity’s adherence cost (RPE/enjoyment -> sustainability).

DEFERRED: aquatic exercise -> function older 2016 (10.1093/ageing/afw102); walking-only -> function/ falls 2024 (10.1123/japa.2023-0183) + walking+health older 2026 (10.1177/08901171261438402); exercise-referral HTA reports 2011/2015 (10.3310/hta15440, 10.3310/hta19600, E-echoes of bmj.d6462).

REJECTED: adherence-predictors cancer SRs (off-stratum); FES-cycling / weight-cycling (off-topic title collisions); Achilles biomechanical-alteration SRs (gait mechanics, not load-progression).

ATTRIBUTED ZEROS: swimming/rowing/calisthenics/bone-stress-”injury”/beginner-runner/running-progression SRs all 0 -> over-constrained or studied under different titles (type/intensity not implement; “stress fracture” not “bone stress injury”); the broader queries that hit cover the cells. Not literature absence.

(4) Acquisition-requested. Fetch attempts: Böhm (prior, OA) -> inbox/. fcvm/T&F/BMJ OA-flagged but PDF-fetch FAILED (publisher walls) -> all maintainer-drop. Net this whole pass: 1 in inbox, ~11 maintainer-drop DOIs handed to the maintainer (list in the 2026-10-04 worklist block). No downgrade, no substitution. Ingest path stays delegated (never main-agent).

[2026-10-04] addendum 2 — tissue-conditioning PROGRAMMING (“build the tissue base”) + “impact-error” construct

Maintainer request: sources for the PROGRAMMING of tissue conditioning (how to progressively load tendon/ bone to tolerance), distinct from adaptation-mechanism + injury-risk-factors. Plus: “impact-error risk” is a maintainer/assistant COINAGE, not a literature term — the real construct is training-error / sudden load increase (“too much too soon”).

SELECTED (worklist 2026-10-04 “Tissue-conditioning PROGRAMMING” block):

  • Nielsen — Training errors & RRI SR 2012 (PMID 22971599, PMC3435910, OA) — THE landmark behind “impact-error”; training-error as dominant RRI cause. maintainer-drop (PMC free).
  • Clifford — Isometric exercise tendinopathy SR+MA 2020 (10.1136/bmjsem-2020-000760) -> FETCHED inbox.
  • Prudêncio — Eccentric Achilles tendinopathy MA 2023 (10.1186/s13102-023-00618-2) -> FETCHED inbox.
  • HSR vs ecc vs iso mixed MA 2024 (10.1016/j.heliyon.2024.e39171) — maintainer-drop.
  • ACWR -> injury SR 2020 (10.1007/s40279-020-01308-6) — maintainer-drop; CONTESTED metric, held with the U-shape-artifact caveat (hold the principle, not the ratio).
  • Exercise loading -> BMD MA 2025 (10.3390/jcm14124109) — maintainer-drop (bone-conditioning arm).
  • Nilsson eccentric Achilles SR 2014 (10.1111/sms.12208) — NOTED optional (superseded by Prudêncio).

CAVEATS recorded on the cell: (1) TRANSFER — tendon-loading evidence is tendinopathy TREATMENT, not healthy-tissue PRECONDITIONING; the naive-tissue-conditioning-in-deconditioned-beginner RCT is a named gap (return-to-running / progressive-tendon-loading-rehab / graded-progression SRs all 0). (2) ACWR contested.

(4) Acquisition. 2 fetched this batch (Clifford, Prudêncio) -> inbox (total inbox now 4: +Böhm, Lazarczuk). 5 maintainer-drop. No downgrade. Ingest stays delegated.

[2026-10-04] addendum 3 — low-floor ENTRY levers (gap-sweep for sedentary/obese cells missed)

Maintainer gap-check: conditioning/sedentary/obese cells not yet captured. Swept the sub-exercise / initiation levers.

SELECTED (worklist “Low-floor ENTRY levers” block):

  • Buffey — interrupting prolonged sitting MA 2022 (10.1007/s40279-022-01649-4) -> FETCHED. New cell (held Sedentary page = Patterson sedentary-TIME, not sitting-INTERRUPTION).
  • Chastin — light-intensity PA & cardiometabolic MA 2018 (10.1136/bjsports-2017-097563, 504c) -> FETCHED. F-check vs held Ekelund any-intensity.
  • Gómez-Redondo — supervised vs unsupervised exercise -> function MA 2024 (10.1007/s40279-024-02024-1) -> FETCHED. Delivery/adherence arm.
  • Rodríguez — exercise snacks MA 2025 (10.1136/bjsports-2025-110027, BJSM) -> fetch returned a landing page (removed from inbox); maintainer-drop/refetch.
  • sms.70114 exercise-snacks MA 2025 + nursrep15030089 RT-in-sarcopenic-obesity 2025 -> maintainer-drop (fetch failed).

ATTRIBUTED ZEROS: “breaking up sitting glucose” / “physical activity snacks fitness” / “stair climbing fitness” all 0 -> reworded hits via “interrupting prolonged sitting” / “exercise snacks” (stair-snacks are a subset). Not absence.

(4) Acquisition. 3 fetched good (Buffey, Chastin, Gómez-Redondo); 1 landing-page removed (Rodríguez); 3 maintainer-drop. Inbox now 7 good PDFs. No downgrade. Ingest delegated.

[2026-10-06] SELECTION RECORD — discover sweep 6 (full --refresh + Engine G + AWAITS + query-driven gap)

(1) Queries. discover.py --refresh --write (engines A-F + coverage_matrix; auto-logged per engine) · acq_contradiction.py --top-rows 12 --write (auto-logged above) · OpenAlex abstract pulls (direct API works/doi: for 14 Engine-G DOIs — NOT auto-logged, hand-logged here) · openalex search x3 + resolve x1 (auto-logged above): %1RM-vs-reps MA (driven by a maintainer query on the Minimal Exercise card) and the Goh exercise-types NMA (resolving an OPEN [AWAITS]). No web search.

(2) Results. Engine A 467 novel · Engine E 546 novel · Engine G 48 unheld over 10 CLOSED rows · GONE 50. Zero-hit attributed: Engine G row 10 (SSB->adiposity/disease) = multi-word title.search over-constraint, a tool artifact (6th sweep running), not literature absence. srcgrep corpus check for the query (Epley / Brzycki / reps in reserve / repetitions to failure / predicted|estimated 1RM, 1,285 files): 0 relevant hits (4 surname false positives in reference lists) -> the %1RM gauge is a genuine fabric gap. The Goh search’s top hit 10.1186/s13643-016-0321-6 is the 2016 PROTOCOL; the review is 10.1007/s40279-019-01082-0.

(3) Adjudication.

  • SELECTED: Goh NMA 2019 10.1007/s40279-019-01082-0 (AWAITS fill; mode ranking) · Li 2024 MR 10.1093/eurjpc/zwae117 (type-D candidate vs held red-meat association; fills the named no-MR hole; weak-instrument caveat) · Nuzzo 2023 10.1007/s40279-023-01937-7 (%1RM<->reps; sources a † gauge on the Minimal Exercise card) · Lean 2019 DiRECT 24-mo 10.1016/s2213-8587(19)30068-3 (F on held 12-mo) · Armitage/CTT 2019 10.1016/s0140-6736(18)31942-1 (statins x age stratum) · Laranjo 2020 10.1136/bjsports-2020-102892 (trackers -> PA, adherence) · Reynolds 2020 10.1371/journal.pmed.1003053 (fibre/whole grain in diabetes, stratum F).
  • DEFERRED: Hill/O’Connor 2022 umbrella 10.1080/10408398.2022.2123778 (restates held NutriRECS pole; shares an author with held OConnor 2017) · Delgado-Lista CORDIOPREV 2022 (established-CHD edge) · Prince 2020 self-report-vs-device PA (check held accelerometer pages first) · Syn 2021 bariatric-surgery mortality (paywall; surgery-comparator for severe obesity, low priority).
  • REJECTED (Engine A/E top-25 slices, by title + registry novelty check): Neuenschwander 2019 + Bellou 2018 T2D umbrellas (echo of held Szczerba 2023 umbrella anchor) · Yu 2020 AD-prevention umbrella (echo of held Livingston 2024 + Zhang 2022) · Wallace 2019 fruit/veg (narrative) · Solmi 2021 mental-disorder onset, Petermann-Rocha 2021 sarcopenia prevalence, Liu 2022 heat exposure, Stevens 2022 prevalence estimates, Emmons-Bell 2022 HF epidemiology (descriptive/non-modifiable, no decision) · Friedenreich 2019 PA in cancer survivors (treatment-of-disease stratum) · García-Hermoso 2018 strength-predicts-mortality (echo of held grip/strength mortality pages) · Fusar-Poli 2018 / Papatheodorou 2019 umbrella-method papers (method corpus CLOSED) · semaglutide/tirzepatide single RCTs (Davies, Garvey, Kosiborod, Butler, Knop; per-agent trials — class-level efficacy already held, per-agent = frontier). Engine G: rows 03/08/11/12 CONCUR; 01/02/04/06/07 DIFFERENT-QUESTION (detail: contradiction-hunt.md).
  • NOT ADJUDICATED: the remaining ~950 Engine A/E rows (below the top-25 cites slice). Unread, not rejected.

(4) Acquisition. Nothing fetched (Discover does not acquire). OA -> fetch candidates for a research pass: Goh 2019, Nuzzo 2023, Lean 2019, Armitage 2019, Laranjo 2020, Reynolds 2020. Maintainer-drop (paywall): Li 2024 10.1093/eurjpc/zwae117. Oliveira (affective response) already STAGED -> ingest-queue.

  • [2026-10-06] openalex fetch “10.1007/s40279-019-01082-0” -> 0 hits: (none)
  • [2026-10-06] openalex fetch “10.1007/s40279-023-01937-7” -> 0 hits: (none)
  • [2026-10-06] openalex fetch “10.1016/s2213-8587(19)30068-3” -> 0 hits: (none)
  • [2026-10-06] openalex fetch “10.1016/s0140-6736(18)31942-1” -> 0 hits: (none)
  • [2026-10-06] openalex fetch “10.1136/bjsports-2020-102892” -> 0 hits: (none)
  • [2026-10-06] openalex fetch “10.1371/journal.pmed.1003053” -> 0 hits: (none)

[2026-10-06] acquisition addendum (field 4) — research on the sweep-6 SELECTED set, at maintainer request. openalex fetch -> inbox/, each PDF checked by page-1 title/DOI: Goh 2019 10.1007/s40279-019-01082-0 (19pp) · Nuzzo 2023 10.1007/s40279-023-01937-7 (19pp; print issue is Sports Med 2024) · Laranjo 2020 10.1136/bjsports-2020-102892 (13pp) · Reynolds 2020 10.1371/journal.pmed.1003053 (22pp). Blocked -> maintainer-drop, no downgrade: Lean 2019 10.1016/s2213-8587(19)30068-3 (no OA pdf) · Armitage 2019 10.1016/s0140-6736(18)31942-1 (OA-flagged, Lancet HTTP 403) · Li 2024 10.1093/eurjpc/zwae117 (paywall). Next: process inbox.

[2026-10-06] SELECTION RECORD — UKK 2-km walk test + ACSM leg-cycling metabolic equation (maintainer request after the MET-estimation challenge)

Query (1): openalex (auto-logged above): UKK walk test · 2-km walk test · two kilometre walking test · 2-km walking test cardiorespiratory fitness · walk test maximal oxygen uptake prediction · ACSM metabolic equation cycle ergometry · ACSM leg ergometry equation · accuracy ACSM cycle ergometry equation · cycle ergometry equation · metabolic equations oxygen uptake cycle ergometer · submaximal cycle test VO2max prediction · field tests cardiorespiratory fitness · walk tests cardiorespiratory fitness · submaximal exercise tests VO2max; SR/MA-filtered: cardiorespiratory fitness estimation, submaximal tests maximal oxygen uptake, walking test. OpenAlex API author/OA lookups by DOI; figshare API for the Kokkinos green copy. Results (2): ACSM metabolic equation cycle ergometry = 0 (title-strictness artifact: the equation is named “cycle ergometry equation” in titles, and that query returned 8). submaximal tests maximal oxygen uptake +SR = 0 (no SR on submax-test validity in titles; the MA filter returned only the RPE-regulated MA, Coquart 2016). No SR/MA specific to the 2-km walk test was found. The UKK literature is the 1991 original plus single validation studies and conference abstracts. Adjudication (3):

  • SELECTED Kokkinos 2018 (10.1177/2047487318772667, EJPC, FRIEND registry, n=6,822 CPX on cycle). It derives a cycle VO2max equation from measured VO2 and compares it to the ACSM leg-ergometry equation (ACSM relative bias 15.46% vs 0.51%). It answers the ACSM-equation request with accuracy data, which the ACSM book alone would not give. It links to held Ross/FRIEND. Tier: large registry cohort, prediction-equation design.
  • SELECTED Oja 1991 (10.1055/s-2007-1024694, Int J Sports Med): the UKK test’s original derivation/validation, the canonical citation for the equation. Method-tier single validation study; needs a source-tier-overrides entry at ingest (the measurement tool itself; no SR exists).
  • SELECTED Marín-Jiménez 2023 (10.1016/j.jsams.2023.03.005, JSAMS): the recent criterion validity AND test-retest reliability of the 2-km walk test. Reliability is what the change-tracking use needs.
  • SELECTED-ADJACENT Mayorga-Vega 2016 (10.1371/journal.pone.0151671, PLoS ONE, psychometric MA, 123 studies). Gold design for walk/run field-test validity (rp 0.42-0.79; 1-mile and 12-min included), but it does NOT cover the 2-km/UKK test (srcgrep-equivalent PDF check: “2-km”/“2,000”/“UKK” = 0 hits, positive control “walk test” = 8). It reports correlations only, not SEE. It is kept as the class-level gold anchor.
  • DEFERRED Lang 1992 (10.1249/00005768-199202000-00019, MSSE): ACSM cycle-equation accuracy, n small, and superseded in scope by Kokkinos. Acquire only if Kokkinos proves insufficient.
  • DEFERRED Laukkanen 1992 (Int J Obes, no DOI): 2-km walk validity in overweight adults. A stratum-relevant extension of Oja; second wave.
  • REJECTED: 6MWT-based prediction papers (clinical/pulmonary/pediatric populations, off-stratum); the ACSM Guidelines book (copyrighted, the equation is reproduced and tested in Kokkinos); the 2002 ACSM-revision conference abstracts (abstract-only); the wearable free-living CRF MA (2025, a different question). Acquisition (4): fetched to inbox/: Mayorga-Vega 2016 (OA, PLoS); Kokkinos 2018 (accepted manuscript via figshare 10753589, not the version of record). BLOCKED -> maintainer-drop: Oja 1991 (Thieme, paywalled); Marín-Jiménez 2023 (Elsevier, openalex says no OA pdf).
  • [2026-10-06] acq-engine G contradiction-hunt [rows=13,14,15,16,17,18,19,20,21,22 per-row=8] -> closed-scanned=10 unheld-candidates=58 zero-yield=2 | top: 10.1111/ijpo.70154; 10.1007/s11154-026-10086-x; 10.1007/s11357-026-02525-z; 10.3389/fnut.2026.1925019; 10.3390/nu18172866 pool-tsv=/workspaces/nutrition/wiki/meta/acq-contradiction.tsv
  • [2026-10-06] openalex search “Mediterranean diet cardiovascular randomized meta-analysis” -> 6 hits: https://doi.org/10.1080/10408398.2019.1565281; https://doi.org/10.1016/j.cpcardiol.2024.102509; https://doi.org/10.1080/27697061.2024.2440051; https://doi.org/10.1093/ehjci/ehaa946.2856; https://doi.org/10.1161/hyp.82.suppl_1.tac134
  • [2026-10-06] openalex search “Mediterranean-style diet prevention cardiovascular disease systematic review” -> 0 hits: (none)
  • [2026-10-06] openalex search “lifestyle intervention metformin diabetes prevention meta-analysis” -> 2 hits: https://doi.org/10.1186/s13098-024-01504-8; https://doi.org/10.2337/db23-659-p
  • [2026-10-06] openalex search “vegetarian diet mortality meta-analysis” -> 2 hits: https://doi.org/10.1016/j.ijcard.2014.07.080; https://doi.org/10.1016/j.ajpc.2021.100182
  • [2026-10-06] openalex search “diet quality score mortality meta-analysis” -> 0 hits: (none)
  • [2026-10-06] openalex search “healthy eating index mortality meta-analysis” -> 1 hits: https://doi.org/10.1016/j.advnut.2023.100166

[2026-10-06] SELECTION RECORD — discover sweep 7 (Engine G CLOSED rows 13-22)

(1) Queries. acq_contradiction.py --rows 13..22 --write (auto-logged) · openalex search x4, all auto-logged above: “Mediterranean diet cardiovascular randomized” MA >=2018, “Mediterranean-style diet prevention cardiovascular disease” SR, “lifestyle intervention metformin diabetes prevention” MA >=2015, “vegetarian diet mortality” MA >=2013, “diet quality score mortality” MA >=2023, “healthy eating index mortality” MA >=2022 · discover.py --novelty on the Cochrane DOI · direct OpenAlex works/doi: abstract pulls for 13 DOIs (hand-logged here, not auto-logged). No web search. The Engine A/E pools are essentially unchanged since sweep 6 this morning, so their top-25 slices were not re-adjudicated.

(2) Results. Engine G: 58 unheld candidates over 10 CLOSED rows. Zero hits: rows 16 and 18 (the same title.search over-constraint, a tool artifact), “Mediterranean-style diet prevention CVD” SR (0; Cochrane found via a direct DOI instead), and “diet quality score mortality” MA (0; the HEI phrasing gave 1). Hand queries: Mediterranean RCT MA 6 hits, lifestyle-vs-metformin 2, vegetarian 2, HEI 1. No-abstract in OpenAlex: Filippou 2020, cpcardiol 2024, Kwok 2014, jclinepi 2017, soard 2025.

(3) Adjudication. Detail per row: contradiction-hunt.md sweep 7.

  • SELECTED: Rees, Cochrane 2019 10.1002/14651858.CD009825.pub3 (gold upgrade over the PREDIMED single-RCT incumbent on row 13; certainty diverges, direction concurs) · Filippou 2020 10.1093/advances/nmaa041 (DASH->BP upgrade over Siervo 2015).
  • DEFERRED: cpcardiol 2024 10.1016/j.cpcardiol.2024.102509 (paywall, no abstract; Rees covers the cell) · soard 2025 10.1016/j.soard.2025.11.024 (surgery vs GLP-1, a comparator question).
  • REJECTED: 27697061 2025 Mediterranean-in-obesity MA (its only hard outcome is PREDIMED) · Kwok 2014 Adventist MA (overlaps held AHS-2) · ajpc 2021 vegetarian and HEI-2015 MAs (concur, read in sweep 4) · jclinepi 2017 GBMC critique (commentary; position held) · Cureus 2023 metformin-vs-lifestyle (5 studies, underpowered null on the held side) · s13098 2024 and jcm 2025 (add-on and drug-vs-placebo questions) · rows 14/17/20 (off-outcome or surrogate endpoints).

(4) Acquisition. Nothing fetched (Discover does not acquire). OA fetch candidates for a research pass: Rees 2019 and Filippou 2020. Maintainer-drop if wanted: soard 2025.

[2026-10-07] SELECTION RECORD — discover sweep 8 (Engine G CLOSED rows 23-34)

(1) Queries. acq_contradiction.py --rows 23,...,34 --write (auto-logged) · 49 openalex search hand queries by the reader agent, all auto-logged above (GLP-1 withdrawal, weight-loss maintenance, weight loss -> mortality, GLP-1 kidney/CV, tirzepatide, LDL per-mmol, SCORE2/PREVENT, risk prediction in older adults, aspirin) · discover.py --novelty on five DOIs · direct OpenAlex works/doi: abstract pulls for about 25 DOIs by the agent and 8 re-pulled by the parent (hand-logged here, not auto-logged). No web search.

(2) Results. Engine G: 82 unheld candidates over 11 rows; row 26 zero (the engine’s query). Ten hand queries returned zero, listed above with -> 0 hits. Read as title.search strictness: rephrasings found hits on rows 23, 24, 26 and 29; the three zero rows not rephrased (GLP-1 CV in non-diabetics, the USPSTF aspirin evidence report, LDL-intensity mortality) were covered by broader queries on the same rows.

(3) Adjudication. Detail per row: contradiction-hunt.md sweep 8.

  • SELECTED: West 2026 BMJ 10.1136/bmj-2025-085304 (primary behind a held second-hand figure) · Budini 2026 10.1016/j.eclinm.2026.103796 (potential disagreer on the extrapolated regain tail) · Nordmo 2019 10.1111/obr.12949 (attenuates Franz under a no-continued-intervention scope) · Neuen 2026 10.1038/s41591-026-04437-z (SCORE2/PREVENT external calibration).
  • DEFERRED: Ganjali 2025 10.1136/heartjnl-2025-325665 (paywall; fills the >=70 calibration gap) · Aronne SURMOUNT-4 10.1001/jama.2023.24945 (concurs with STEP-1; echo) · Kow 2026 10.1002/edm2.70325 (0 cites; same linear projection as West).
  • REJECTED: Byrne 2022 10.1001/jamainternmed.2022.0134 (different quantity; underpowered “inconclusive”) · Navarese 2018 10.1001/jama.2018.2525 (per-trial RRs not per-mmol; different question) · Silverman 2016 10.1001/jama.2016.13985 (concurs, overlapping trials) · Singh 2019, Barte 2010 (concur) · Dombrowski 2014, Hartmann-Boyce 2023, SURMOUNT-5 (different questions) · Chen 2025, Marzano 2026, Müllertz 2024 (concur/echo; Badve already held) · Gelbenegger 2019 (same 13 trials as Zheng) · engine pool remainder (deposits, review letters, off-topic).

(4) Acquisition requested. West, Budini, Nordmo, Neuen folded into acquisition-worklist.md (all OA, not yet fetched; route through research/process inbox). Ganjali left as a maintainer-drop option.

[2026-10-21] SELECTION RECORD — discover sweep 9 (refresh + PA/fat-loss gap search + Engine G rows 35-46)

(1) Queries. discover.py --refresh --write (engines A-F + coverage_matrix regenerated, --no-log) · acq_contradiction.py --rows 35,...,46 --write (auto-logged) · parent openalex search gap queries (auto- logged above): balance-training dose, exercise -> falls, rest interval, warm-up, injury-prevention programmes, self-weighing, weight-loss rate, exercise adverse events, pre-participation screening, DOMS · reader-agent hand queries for rows 35-46, 53 in all (auto-logged above; one --type network-meta-analysis usage error, not run) · discover.py --novelty on ~15 DOIs · direct OpenAlex works/doi: abstract pulls (parent ~14, agent ~40) and Europe PMC REST abstracts where OpenAlex had none (Graudal, Radaelli, Mizuta, Huang 2021, Kim 2025, Haze 2025, Forster 2025, Pelland 2025) — hand-logged here. No web search.

(2) Results. Pool delta vs committed: Engine A +5 DOIs, Engine E +31; coverage matrix 43 blanks (was 47, none new). Engine G: 73 candidates over 12 rows, row 45 zero on the engine query. Zero-hit hand queries, all attributed to title.search strictness (each recovered by a rephrasing): “balance training dose-response older adults”, “rest interval between sets”, “warm-up injury prevention”, “rapid versus gradual weight loss”, “adverse events exercise training”, and the agent’s five (row 35 x2, rows 42/43, 44, 45 — listed in contradiction-hunt.md sweep 9).

(3) Adjudication. Detail: acquisition-worklist.md UPDATE 2026-10-21 + contradiction-hunt.md sweep 9.

  • SELECTED (gap search): Sherrington 2017 10.1136/bjsports-2016-096547 (falls dose: >3 h/week of balance-challenging exercise) · Madigan 2015 10.1186/s12966-015-0267-4 (self-weighing RCT-MA) · Zheng 2014 10.1002/oby.20946 (self-weighing harms) · Turicchi 2019 10.1111/obr.12849 (rate of loss vs regain) · Ekelund 2026 10.1016/s0140-6736(25)02219-6 (small PA changes, paywall).
  • SELECTED (Engine G): Graudal 2020, Radaelli 2024, Wang 2025, Filippini 2020, Brand 2022, Rahimi 2021, Lai 2026, Di Federico 2023, Zhang 2026, Han 2022 / Lang 2024, Nazarzadeh 2022.
  • DEFERRED: Lesinski 2015 10.1007/s40279-015-0375-y (surrogate balance outcome; behind Sherrington 2017) · Yang 2025, Khalafi 2025, Yao 2023 (pool delta) · Grönroos 2024 · Lu 2026 (no abstract) · Weeldreyer 2024.
  • HOLD: Pelland 2025 (cut as echo; maintainer call to reverse).
  • REJECTED: rest-interval golds Grgic 2017 10.1007/s40279-017-0788-x and Singer 2024 10.3389/fspor.2024.1429789 (trained-lifter hypertrophy: frontier zone) · warm-up golds Fradkin 2010 10.1519/jsc.0b013e3181c643a0, Ding 2022 10.3390/ijerph19106336, football/Nordic programmes (sport- specific; weak transport to untrained adults; Lauersen held) · exercise-ECG screening 2016 10.1007/s40279-016-0501-5 (different question) · DOMS-treatment MAs (do not test a stop rule) · pool-delta noise (Noubiap, Ge, Wattanapisit, Tsaban, Liang, Zhai, Arjmand, Reynolds, 20 Zhang-umbrella coupling hits) · Engine G concurring/different-question candidates listed in the sweep-9 entry · García 2023 (already cut).
  • NAMED GAPS (expectancy fails; labelled practice default is the route): pre-exercise screening, pain-vs- soreness stop rule, technique learning, scheduling, rest between sets for novices, warm-up for novices.

(4) Acquisition requested. None fetched in this op. 14 OA survivors queued in the worklist for research -> process inbox; maintainer-drop: Ekelund 2026, Han 2022.

Maintainer asked to fetch the sweep-9 OA survivors and, for the first-week practicalities, to find the sources that well-established protocols must have (“there are studies for rest between sets, warm-up, soreness”). The sweep-9 verdict that these were named gaps (no gold plausible) is REVISED: it judged transport (trained-lifter / sport samples) instead of first asking what exists, and it missed a source already in cut/ (ACSM 2026) that covers rest and time of day directly.

(1) Query. openalex fetch x15 (sweep-9 OA list + Lang 2024 as Han’s OA route); Europe PMC REST query=DOI:"<doi>" (hand-logged) for PMCIDs of the failed fetches; europepmc.org/articles/<PMCID>?pdf=render and pmc.ncbi.nlm.nih.gov/articles/<PMCID>/pdf/ via curl, fetch_browser.py local, and --service steel (hand-logged). openalex search (auto-logged above): rest interval RT (SR) · rest between sets (SR) · warm-up injury (SR) · warm-up performance (MA) · delayed onset muscle soreness (SR) · stretching muscle soreness · post-exercise recovery techniques (MA) · painful exercises chronic musculoskeletal pain · preparticipation health screening · exercise acute cardiovascular events · supervised unsupervised RT (MA) · supervised versus unsupervised training older adults · implementation intentions PA (MA) · rate of weight loss randomised · progression models RT · quantity and quality of exercise · RT injury (SR) · PAR-Q · supervised unsupervised balance strength (SR) · supervision RT · warming up prevent injury · rapid gradual weight loss (MA). srcgrep-equivalent pdftotext scan of the cut ACSM 2026 binary (rest/warm/soreness/ technique/supervis/screen/novice).

(2) Results. Zero-hit: rest between sets (SR) and supervised unsupervised resistance training (MA) — both title-strictness artifacts (the looser rest interval and supervised ... balance muscle strength queries found the golds). Fetch failures are tool/publisher artifacts, not absence: Wiley, Cochrane, AHA, Lancet, MDPI, BMJ return 403 or an HTML page; openalex’s OA flag was wrong for 6 DOIs (no OA PDF); Europe PMC render and PMC both serve a reCAPTCHA page; Steel returned HTTP 402 (no paid plan). ACSM 2026 text: short (<1 min) vs long (>1 min) between-set rest and morning vs evening training did not affect strength; inter-set rest has 2 reviews (n=982) for strength; 0 hits for warm-up and soreness.

(3) Adjudication.

  • SELECTED (protocol sources): ACSM 2026 (RESURFACED from cut/ — its resurface condition, “prescription specifics (sets/rest/freq) beyond the NMA”, is now met; covers rest, time of day, safety) · Garber/ACSM 2011 position stand (warm-up, cool-down, progression, screening; the recommendation-tier protocol source) · Ding 2022 warm-up programmes -> sport injury MA (newest pooled warm-up/injury MA; supersedes Herman 2012 and Fradkin 2006 for injury) · Fradkin 2010 warm-up -> performance SR-MA · Herbert 2002 BMJ stretching before/after exercise -> soreness and injury SR · Herbert Cochrane 2011 stretching -> soreness (supersedes the 2002 BMJ on soreness; take both, the BMJ carries the injury arm) · Dupuy 2018 recovery techniques -> DOMS/fatigue MA · Smith 2017 BJSM painful vs pain-free exercise MA (the pain stop rule) · Riebe 2015 ACSM preparticipation screening (consensus; the instrument the deliverable would cite) · Franklin 2020 AHA exercise-related acute CV events (warning symptoms; risk sizing for a sedentary starter) · Silva 2018 implementation intentions -> PA, MA of RCTs (scheduling) · Lacroix 2017 supervised vs unsupervised training in older adults MA (technique / supervision) · Purcell 2014 Lancet D&E rate-of-loss RCT (direct test of the question Turicchi answers by between-study regression).
  • DEFERRED: Grgic 2017/2018 rest-interval SRs (hypertrophy/strength in mostly trained samples; ACSM 2026 already pools them) · Herman 2012 and Fradkin 2006 (superseded by Ding 2022) · Thompson 2007 AHA (superseded by Franklin 2020) · ACSM 2009 progression stand (superseded by ACSM 2026) · PAR-Q+ Warburton 2011 (the instrument; Riebe covers the algorithm) · Ashtary-Larky 2020 rapid vs gradual loss -> body composition/RMR MA (surrogates; behind Purcell) · Lesinski 2015 (unchanged, behind Sherrington).
  • REJECTED: the sport-specific FIFA 11+ RCTs (Soligard 2008, youth football) — off-stratum for an untrained adult; the MA already pools them.
  • Pelland 2025 HOLD -> cut stands (maintainer delegated the call): hypertrophy/strength surrogates in a young, mostly male, partly trained sample sit on the frontier side of the exercise-programming taper; Radaelli 2024 (older-adult function) and ACSM 2026 (variable-level, any training status) cover the decision-bearing cells.

(4) Acquisition. Fetched to inbox/ (8): Sherrington 2017 · Madigan 2015 · Turicchi 2019 · Lang 2024 · Dupuy 2018 · Smith 2017 · Silva 2018 · ACSM 2026 (copied from cut/). Maintainer-drop (blocked or no OA PDF), DOIs in acquisition-worklist.md UPDATE 2026-10-21 (b).

Addendum — second fetch pass (hand-logged). Europe PMC rest/<PMCID>/fullTextXML: HTTP 200 for PMC7429027, PMC9363242, PMC8473559, PMC12911277, PMC10510850, PMC9622419, PMC9140806 (saved as JATS); HTTP 500 (not in the OA XML subset) for PMC8094404, PMC12458985, PMC13397323, PMC119442. Those four were (REVERTED: maintainer ruled out paper HTML, files deleted, back to maintainer-drop) saved as PMC article HTML via fetch_browser.py --html (full text checked by pandoc word count: 96.7k / 14.7k / 10.7k / 5.0k words; no CAPTCHA markup). Europe PMC DOI lookup found no PMCID for Garber, Riebe, Franklin, Fradkin, Herbert 2011, Lacroix, Purcell, Ekelund 2026, Han, Zheng, Radaelli. Playwright on publisher PDFs: Springer (Lacroix, Radaelli) serves a JavaScript bot check, Wiley (Zheng) and LWW (Garber) return 403. Exa was not used to acquire: it returns page text into the model’s context, and writing that text to a file would be a model reproduction of the source, not a copy.

  • [2026-10-22] openalex search “Adverse events of exercise therapy in randomised controlled trials” -> 1 hits: https://doi.org/10.1136/bjsports-2018-100461
  • [2026-10-22] openalex fetch “10.1136/bjsports-2018-100461” -> 0 hits: (none)

[2026-10-22] SELECTION RECORD — Gate 55 corrective 5 (Niemeijer 2020, Pelland 2025)

(1) Query. OpenAlex title search «Adverse events of exercise therapy in randomised controlled trials» (from 2019); OpenAlex fetch on the DOI; Europe PMC REST DOI:10.1136/bjsports-2018-100461; fetch_browser.py --download on the BMJ full.pdf URL, local Chromium then --service steel. (2) Results. OpenAlex 1 hit, Niemeijer BJSM 2020 (10.1136/bjsports-2018-100461, oa=True, 156 cites). OpenAlex fetch and local Chromium both saved the BMJ HTML landing page, not a PDF (deleted; paper HTML is not acquired). Europe PMC: PMID 31563884, no PMCID, not OA there. Steel: HTTP 402 (payment required). (3) Adjudication. Niemeijer SELECTED: gold-design SR+MA of adverse events across exercise-therapy RCTs (RT subgroup), fills the [AWAITS (Niemeijer - Exercise Therapy Adverse Events 2020)] safety cell on the RT prescription page and the spec gap “general adverse events of exercise”. Pelland 2025 NOT re-requested: the 2026-10-21 cut stands (rejected-sources.md row; frontier-side surrogate sample). Gate 55 corrective 5 carried the item forward from Gate 54 without noting that cut; flagged to the maintainer, cut not overridden. (4) Acquisition. Niemeijer: blocked -> maintainer-drop, DOI 10.1136/bjsports-2018-100461.

[2026-10-26] SELECTION RECORD — DOI resolution for the 8 OPEN [AWAITS] tokens (maintainer request)

  • (1)+(2) Channel OpenAlex search (title.search) + per-DOI works/doi: author check. First pass of 5 descriptive queries returned 0 hits each: a title.search strictness artifact (long multi-term strings), not literature absence. Second pass with short title phrases hit all 5 (lines above). Pelland + Niemeijer DOIs were already on file (worklist).
  • (3) Adjudication: each DOI SELECTED as the registry row’s paper (title + first three authors match the placeholder row): Fink 2009 10.1016/j.eururo.2009.03.031 (OA) · Hakam 2024 10.1001/jamanetworkopen.2024.47621 (OA) · Rhee 2016 10.1016/j.ijcard.2016.04.109 (paywall) · Qiu 2022 10.3390/nu14235073 (OA) · Jung 2021 10.1017/s0007114521000088 (paywall) · Niemeijer 10.1136/bjsports-2018-100461 (OA flag, but BMJ blocks fetch; online 2019, slug year 2020) · Pelland 2025 10.1007/s40279-025-02344-w (CUT stands 2026-10-21). REJECTED: the 2009 Eur Urol editorial comment 10.1016/j.eururo.2009.03.032 (not the SR); the other sodium/renin hits (10.2215/cjn.09110914, the SGLT2-inhibitor papers) — different question.
  • (4) Acquisition: none requested this pass. Registry doi cells filled for Fink, Hakam, Niemeijer (Qiu, Jung, Rhee, Pelland already carried theirs); acquisition priority per acquisition-worklist.md (Niemeijer the only one previously approved; the rest unadjudicated).
  • [2026-10-27] openalex resolve “10.7326/M14-0773 10.1056/NEJMoa1600175 10.1056/NEJMoa1600177 10.1038/s41591-023-02286-8 10.1093/ajh/hpu028 10.1056/NEJMoa0801369 10.1016/S0140-6736(02)11600-X” -> 7 hits: https://doi.org/10.7326/m14-0773; https://doi.org/10.1056/nejmoa1600175; https://doi.org/10.1056/nejmoa1600177; https://doi.org/10.1038/s41591-023-02286-8; https://doi.org/10.1093/ajh/hpu028

[2026-10-27] SELECTION RECORD — post-Gate-57 acquisition review of the deferred BP-cluster candidates (maintainer request)

  • (1)+(2) Channel: OpenAlex resolve on 7 DOIs (line above; all 7 resolved) + srcgrep held-coverage checks: DECIDE/Yuan in Lai (Yuan 2023 is Lai ref 69, so DECIDE-Salt is already pooled); HOPE-3/Lonn 0/1 files in the BPLTTC 2021 chunk (trial list not in the held chunk, so pool membership unverified, not disproved); Mente 2016 PURE sodium (held) covers the sodium J-curve cohort arm.
  • (3) Adjudication. RECOMMENDED: Lonn - HOPE-3 BP arm 2016 10.1056/NEJMoa1600175 (OA PDF per OpenAlex) — the prespecified within-trial baseline-SBP subgroup in no-CVD intermediate-risk people on modern drugs; tests BPLTTC 2021’s pooled no-interaction reading and the Mente premise the tension page rebuts, and the CLAUDE.md within-study-modification rule; also modern-drug tolerability for the Wang G-gap; supersedes the registry placeholder Yusuf 2016 10.1056/NEJMoa1600177 (factorial combined paper) for the BP question. RECOMMENDED: Sundstrom 2015 10.7326/M14-0773 (paywalled) — the gold that Wang says «remains in conflict» with it; the fabric holds only Wang’s paraphrase (incl. «96% … type 2 diabetes», flagged unverifiable); the disagreer the CLOSED lighthouse row 127 needs; BPLTTC school (Rahimi co-author), so F/D vs Wang, never E vs BPLTTC 2021. DEFERRED (unchanged): Greenwood 2024 (trial-set echo of Lai/Brand), Huang 2024 (SSaSS mechanism split), Rhee 2016 (marker), Qiu 2022 / Jung 2021. REJECTED for now: Graudal 2014 10.1093/ajh/hpu028 (cohort U-curve, same group as held Graudal 2020; Mente 2016 + the U-artifact checks already decide the arm); Graudal 2015b (race within-study subgroup; Brand + Graudal 2020 ethnicity subgroups held); HYVET 10.1056/NEJMoa0801369 (inside the staged Rahimi 2021 age-pool; revisit only if Rahimi’s >=80 band is thin); PROSPER 10.1016/S0140-6736(02)11600-X (inside held CTT 2019 older-people); SCORE2-OP (cut stands); DECIDE-Salt 10.1038/s41591-023-02286-8 (pooled in Lai; adds care-home safety only — optional).
  • (4) Acquisition: none fetched; two recommended to the maintainer (Lonn OA-fetchable, Sundstrom maintainer-drop). awaits-ledger rows stay DEFERRED until the maintainer decides.
  • [2026-10-27] openalex fetch “10.1056/NEJMoa1600175” -> 0 hits: (none)
  • [2026-10-27] (4) update: Lonn 2016 fetch FAILED (openalex fetch HTTPError; fetch_browser local Playwright HTTP 403 on nejm.org PDF; Europe PMC search 503) -> maintainer-drop, with Sundstrom 2015.
  • [2026-11-02] openalex search “Sauna bathing is associated with reduced cardiovascular mortality and improves risk prediction in men and women” -> 1 hits: https://doi.org/10.1186/s12916-018-1198-0
  • [2026-11-02] openalex fetch “10.1186/s12916-018-1198-0” -> 0 hits: (none)