Research Synthesis: Cognition Durations

Artifact

Living evidence brief from agent-v3-full-paper-live

Reviewer panel scores

Review verdicts

Why

Review decision

To resubmit, address

1. Resolve the misattribution between 'Association of Objective Sleep 2023' (Alzheimers Dement. conference abstract, R12) and Tang 2024 (Brain Communications, the actual full study of objective sleep duration, cognition, and brain aging biomarkers). Either rename the cited source or revise the description to match the conference-abstract status and directness tier.
2. Recast Schrenk 2023 throughout as a published protocol with no empirical results, not as a direct-anchor RCT. Remove the claim that it provides a 'direct mechanistic/biomarker' human evidence anchor and instead note it as a planned study whose results would address the mechanism-to-clinic gap.
3. Correct the description of Sampaio 2020: replace '17-year longitudinal design' and 'large cohort of men and women' with the cross-sectional design and n=102 reported in the bundle excerpt.
4. Either remove or integrate the Yang 2026 positive effect sizes (SMD 0.58–0.88) into the dosing-PK narrative as a direct counter-signal to the Asteasu 2024 negative inpatient finding, or explicitly justify why this Bayesian NMA is not pooled into the synthesis.
5. Enumerate the 22 cross-study disagreements in a named Tensions and Gaps table or list, mapping each to the specific source pair and outcome class, rather than asserting the count in the abstract.
6. Add a one-paragraph operational definition of 'Cognition Durations' clarifying which types of duration (sleep, exercise, intervention, follow-up) fall in and out of scope.
7. Verify or replace the Ioannidis 2005 surrogate-endpoint citation with a bundle-grounded source, or remove the citation.
8. Add a sentence in the Limitations section explicitly noting that Yang 2026's positive community-dwelling effects and Asteasu 2024's negative inpatient effects represent a directional tension that the current synthesis does not adjudicate.

Major issues

• Internal inconsistency: 'Association of Objective Sleep 2023' is cited as both a systematic review in the Cognitive section and used to anchor a 6–8h reference band, but the bundle entry is an Alzheimers Dement. 2023 conference abstract (alz.075234, R12) — this is not a full systematic review and is treated inconsistently as 'review-level' elsewhere. The Tang 2024 Brain Communications paper (doi 10.1093/braincomms/fcae144) is the actual systematic study of objective sleep duration with cognition and brain aging biomarkers; the manuscript misattributes this body of work to the conference abstract.
• The Schrenk 2023 source is described in the manuscript as 'the only direct human trial in the corpus' for contextual-other outcomes, but its own bundle excerpt identifies it as a protocol with no empirical results yet. The manuscript simultaneously calls it a 'direct mechanistic/biomarker RCT' providing a 'direct anchor' — this overstates evidence that does not yet exist.
• Source count and directness classification inconsistency: abstract states '2 direct clinical sources, 11 adjacent clinical sources'; the contextual-other table lists Schrenk 2023 as the sole direct anchor on that outcome class, but the Findings Map table also lists '1 direct' under Contextual Adjacent Evidence. Asteasu 2024 is the other direct clinical source. This is internally consistent numerically but the prose treatment of Schrenk 2023 as a results-bearing direct trial contradicts its protocol status.
• The '22 cross-study disagreements' figure is asserted but never enumerated or mapped; no explicit Tensions and Gaps section surfaces named contradictions, only a generic one-sentence call for future studies.

Minor issues

• The title 'Cognition Durations' is ambiguous and never explicitly defined as a construct; it appears to conflate sleep duration, exercise duration, intervention duration, and follow-up duration without a scoping clarification.
• The Sampaio 2020 description claims a '17-year longitudinal design' but the bundle excerpt describes a cross-sectional study of 102 institutionalized older adults.
• Sampaio 2020 is described in the muscle-function section as reporting 'no primary effect estimate' with 'P > 0.05' and 'unclear direction,' yet earlier the Findings Map tables it as 'mixed signal in 1/1 sources.' The qualitative label 'mixed' is not justified by the bundle text.
• Yang 2026 is cited as 2026 (future-dated) and presents strong positive SMDs (0.58–0.88) for aerobic/resistance/multimodal exercise on cognition; the manuscript does not engage with these positive signals anywhere in the dosing-PK section, which is a selective treatment of a direct tension.
• The 'evidence-tension synthesis' explicitly limits quantitative pooling to ≥3 comparable endpoints; the corpus does not meet this bar for any outcome class, yet the manuscript still presents per-class narrative effect directions as if pooled.
• Population specificity (ambulatory older adults, exercise trials) is named in Limitations but is not threaded back into the Findings Map to constrain interpretation.
• The Ioannidis 2005 surrogate-endpoint caution is cited but Ioannidis 2005 does not appear in the source bundle, so its grounding cannot be verified.

Reviewer note

This evidence map covers a heterogeneous corpus on cognition-relevant durations (sleep, exercise, intervention) in older adults, with 13 admitted sources spanning RCTs, observational cohorts, reviews, and a protocol. The search summary, PRISMA-ScR framing, admission funnel, and risk-of-bias plan are explicit and auditable, and the abstract correctly declines to issue a single anti-aging claim, which is appropriate for the evidence-map genre. Source attribution is mostly traceable through the bundle, and the Limitations section is concrete (single-source anchors, narrow endpoint coverage, mechanism-to-clinic gap). However, several substantive errors prevent an accept. The 'Association of Objective Sleep 2023' citation is anchored to a conference abstract (alz.075234) rather than the full Tang 2024 Brain Communications paper that actually investigates objective sleep duration, cognition, and brain aging biomarkers; the manuscript treats it as a full systematic review. Schrenk 2023 is labeled a direct mechanistic/biomarker RCT anchor despite being an explicitly described protocol with no results. Sampaio 2020 is described as a 17-year longitudinal cohort when the bundle excerpt describes a cross-sectional n=102 study. Yang 2026's strong positive SMDs for aerobic, resistance, and multimodal exercise on cognition are not integrated into the dosing-PK narrative despite being a directly relevant counter-signal to the Asteasu 2024 inpatient null/negative finding. The 22 cross-study disagreements are asserted numerically but never surfaced or enumerated in a Tensions and Gaps table. The findings map does group evidence by outcome class and avoids collapsing into a single causal claim, which is correct for an evidence map. But the partial misalignment between manuscript text and the cited bundle entries (source identity swaps, overstating a protocol as a results-bearing trial, misrepresenting a study design) means claim-to-source support is only partial. These are bounded fixes — the article type is right, the scope is right, the heterogeneity is mostly mapped — but the corrections touch source identification, evidence directness, and the explicit surfacing of named tensions. Revise, not reject, because the underlying structure is sound and the errors are remediable without a scope reset.

Panel metadata

Models: MiniMax-M3 + google/gemma-4-31b-it + mistralai/mistral-small-2603

Route: consensus

Prompt: reviewer-v11-research-synthesis