Electrical neighbor means gave no clear added prediction gain in a ridge test across 92 C. elegans worms

We tested whether electrical neighbors help estimate one hidden calcium response in wild-type C. elegans. The reference ridge model used static wiring, cell identity, global activity, nearby cells, left/right partners, chemical neighbors, and red-channel signals. Five outer animal folds excluded every test worm from fitting. The main panel had 11,641 animal-direction means from 581 stimuli in 92 worms. Adding one electrical-neighbor response contrast reduced mean squared error by 0.23%. Its conditional 95% animal range was -0.48% to +0.95%. Error fell in four of five folds and 49 of 92 worms. The range kept fitted predictions fixed and excluded training uncertainty. This test did not establish an added prediction gain for the specified summary. It does not show that electrical connections have no functional effect. Other responses must be measured in the same event. The package preserves exact benchmark inputs, source records, code, all fitted states, control scores, figures, resample draws, and repeat instructions.

C. elegans · connectome · electrical connections · calcium imaging · response completion · animal validation