{
  "harness": "sj-harness 0.8.4",
  "job": "job:c1eed4dd1566157d0eb4955b42794e61",
  "kind": "adversarial_review",
  "bundle": "sha256:8308c6629e5531886f7129075efaa432d02b073fcc58ab2817d3e119bc3cf00c",
  "claims": [
    {
      "local_id": "C1",
      "claim_id": "claim:8d3931de5119b5a5b9bedf4a4022b187b5d7025b384c975fb53fcf4791b4d7cf",
      "verdict": "minor_issues",
      "significance": "minor",
      "reason": "The numerical results reproduce, but the gender covariate is all zero because F is compared with female/male labels, and HR and SBP analyses use different samples. Correct-label Docker and independent HC3 sensitivities preserve the primary directions; qualify the conclusion as no predicted association detected rather than demonstrated equivalence or general absence of slowing."
    },
    {
      "local_id": "C2",
      "claim_id": "claim:e932584150ab1a7d45c421ba4e24a89a4e8a74c5fe37833c6ef7c6f630348d3d",
      "verdict": "sound",
      "significance": "minor",
      "reason": "The AUC and nominal one-sided reactivity estimate independently reconstruct and reproduce from raw data. The corrected-label reactivity model remains negative, p=0.0342 one-sided; these are explicitly uncorrected descriptive secondary associations, not evidence that blunting excludes impaired recovery or that depressive symptoms causally improve it."
    }
  ]
}
