{
  "study_id": "NCT04312269",
  "agent": "baseline",
  "run": 3,
  "p_primary_hypothesis_supported": 0.12,
  "effect_direction": "favours intervention",
  "effect_size_90pct_interval": {"low": -0.45, "high": 0.75, "metric": "Cohen's d, between-group difference in WMFT change (time-based score), baseline to 6 weeks, TMR vs sham"},
  "moderator_prediction": "none detectable; if anything SWS-only TMR roughly equal to all-phase, since lab TMR benefits come from NREM2/SWS and REM cueing adds little, but arms of ~17 cannot resolve it",
  "claim_ids_used": [],
  "rationale": "Lab TMR meta-analytic effect is small (g=0.29) and concerns memory of the cued task, not generalized functional arm ability; transfer to WMFT in chronic stroke after a home/overnight protocol should be smaller still, perhaps d~0.1-0.2. With 70 participants split over four arms (~17 per arm), power to detect d=0.2-0.3 in a pairwise contrast is under 15%. Both arms receive identical MyoCI training, so the comparator is strong. WMFT time scores are noisy and skewed. Small single-lab trials sometimes report positive findings, and pooling TMR arms could raise power slightly, but the probability of a significant primary contrast remains low."
}