I23f-RRR-01. RESIDUAL RECOVERY RESERVE (RRR): OPERATIONAL MEASUREMENT AND INDEPENDENT HOLDOUT VALIDATION PROTOCOL

This is the authors' abstract. We don't add key points for this paper.

This methodological record consolidates the operational definitions, eligibility criteria, development-model freeze, and completed held-out evaluation of an electrodermal candidate for residual recovery reserve (RRR). Nine participants formed the development cohort and eighteen different participants from the same public PhysioNet dataset formed the held-out cohort. Model A predicted late behavioral error from the pre-second-stress behavioral error; Model B added the difference in mean electrodermal activity between two recovery windows. Held-out mean absolute error (MAE) was 150.05 for Model A and 150.90 for Model B (B minus A = +0.85). The proposed electrodermal feature did not improve average prediction in this sample. The record is published after inspection of held-out outcomes and is not a prospectively registered study.