The 95% bootstrap interval for reported hand dominance prediction gain from cortical area differences included zero in 427 AOMIC test participants under independent participant sampling

We tested whether 34 normalized left-right cortical area differences improve prediction of reported hand dominance across AOMIC cohorts. We fitted models in 928 ID1000 participants and tested them in 427 PIOP1 and PIOP2 participants. The reference used age, recorded sex, estimated head size, and 34 bilateral mean areas. Adding the differences changed test log loss from 0.35715 to 0.35958 nats per person. The gain was -0.00243 nats. Its 95% bootstrap interval was -0.05716 to 0.01429 nats. Each of 2,000 draws sampled training and test people and fitted both models again. The interval assumes independent people and fixed source cohorts. It included zero, so this procedure did not establish a gain. The result does not establish equivalence or absence of an anatomical relation. A constant training prevalence prediction had lower test loss than either fitted model. All six planned secondary checks are reported. The package contains source bytes, provenance, code, predictions, parameters, resampling indices, figures, and verification records.

hand dominance · cortical area · cortical asymmetry · prediction · AOMIC · bootstrap