Author(s) :

Vivaan Patni.

Volume/Issue :

Abstract :

Bioelectric signals help cells coordinate tissue shape during development and regeneration, but it is unclear whether a hidden pattern can be inferred from voltage correction after damage. This study tested whether a partial simulated voltage movie could reconstruct a hidden target-voltage map and predict one of two regenerative outcome classes. A two-dimensional tissue model contained 508 active positions, and 160 independent baseline trajectories were evaluated across six observation windows and five levels of measurement noise. At 5% noise and 30% observation, inverse-movie accuracy was 95.6% (95% confidence interval [CI], 91.9%–98.8%), area under the receiver-operating-characteristic curve was 0.995, anatomical macro-Dice was 0.861, and normalised root-mean-square error was 0.145. Accuracy exceeded the immediate post-damage snapshot by 0.450 (95% CI, 0.362–0.537), but differed from the same-time snapshot by only 0.006 (95% CI, 0.000–0.019). A separate 160-trajectory holdout cohort with shifted parameter ranges reached 95.0% accuracy. All three declared misspecification conditions also reached 95.0% accuracy. When outcome targets were made partially overlapping and their separation was reduced to 0.15 of the original difference, accuracy fell to 62.5%. Ablations showed that smoothing and coefficient bounds materially affected spatial reconstruction, although binary accuracy was unchanged. Correction and coupling rates remained poorly identifiable despite accurate target prediction. These results support target reconstruction within this declared simulation, but they do not establish recovery of the biological mechanism or a universal voltage-to-anatomy code.