AI Discovery of Mechanisms of Consciousness, Its Disorders, and Their Treatment

Le résumé fourni par la source

This repository includes software and a pre-trained neural network developed at UCLA. © 2025 The Regents of the University of California on behalf of UCLA. The software is made available under the UCLA TDG Academic Software License (non-commercial, share alike). The full license text is included in the distribution.The files contain code and configuration files used in the study “AI Discovery of Mechanisms of Consciousness, Its Disorders, and Their Treatment.” The materials support large-scale neural field simulations of conscious and disordered brain states, as well as machine-learning–based prediction of level of consciousness from cortical signals.Simulations were performed using a modified version of NFTsim, an open-source neural field theory simulator. The modification introduces a scaled dendritic response function that allows the synaptic gain parameter to be specified independently for each connection. Apart from this change, NFTsim’s numerical solvers and architecture remain unchanged. Plain-text configuration files are provided for a baseline conscious brain model and for 280 in silico cases of disorders of consciousness generated by systematic variation of synaptic and connectivity parameters.The repository also includes MATLAB code for predicting level of consciousness from cortical time series data using a trained neural network. The prediction pipeline filters and resamples 10-second cortical signals, predicts a Glasgow Coma Scale (GCS) score in the range 3–15, and maps this to a continuous consciousness index between 0 and 1.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.