{
  "schemaVersion": "gaiaworld.release-state.v1",
  "benchmarkId": "gaiaworld-arc-h1-2025-v1",
  "state": "published",
  "stages": {
    "protocolReady": true,
    "sealed": true,
    "scored": true,
    "published": true
  },
  "evidence": {
    "manifestHash": "07a540e1e60e130547e434ff6300b030768a265ac02edca7535b724ce5a07108",
    "benchmarkHash": "d896bc70b7555a867304b90a6237ade5025176fe556abbfef318e8d966fe00da",
    "sealedAt": "2026-08-30T04:20:52Z",
    "comparisonInvariant": "100 held-out cases x 100 frozen panel dimensions; the same denominator and predeclared scoring contract were used for all three frozen model lanes.",
    "manifestKind": "pre_reveal_scoring_lock",
    "benchmarkHashKind": "semantic_score_hash",
    "scoreFileSha256": "f38b1f9ff029991cf140422b67f05e70da053cb87be7c62517e34c51fa3e9ecf"
  },
  "metrics": {
    "scope": "target-ridge frozen model lane on GaiaWorld's pre-frozen 100-gene panel",
    "caseCoverage": 1,
    "abstentionRate": 0,
    "meanAbsoluteError": 0.13257230776900006,
    "rootMeanSquaredError": 0.20639483736455136,
    "directionalAccuracy": null,
    "replayIntegrity": 1,
    "panelPDSL1MidrankNoExclusion": 0.5332000000000001,
    "officialArcLeaderboardMetrics": false
  },
  "baselines": {
    "no-change": {
      "meanAbsoluteError": 0.07615881768500006,
      "rootMeanSquaredError": 0.14742479246557294,
      "panelPDSL1MidrankNoExclusion": 0.5050000000000001
    },
    "empirical-mean": {
      "meanAbsoluteError": 0.08194923586099996,
      "rootMeanSquaredError": 0.14474121927189892,
      "panelPDSL1MidrankNoExclusion": 0.5049999999999999
    }
  },
  "failures": {
    "primaryMetricOutcome": "negative",
    "targetRidgeCaseWinRateVersusNoChange": 0.11,
    "targetRidgeCaseWinRateVersusEmpiricalMean": 0.08,
    "directEvidenceAbstentions": 100,
    "directEvidenceCoverage": 0
  },
  "resultUrl": "/gaiaworld-experiment-001-result.json",
  "scientificBoundary": "Experiment #001 is a retrospective held-out 100-gene panel benchmark. The frozen target-ridge model failed the predeclared primary panel-MAE comparison against both baselines. A modest secondary panel-discrimination signal does not override that negative primary result. This is not an official Arc leaderboard result and does not establish causal validity, general biological prediction ability, patient benefit, safety, efficacy, or clinical utility."
}