A State of the Art Brain Language Foundation Model
We introduce a large-scale MAE-based Brain Foundation Model trained on diverse EEG data from global cohorts. Using 4D positional encoding and a PatchConv frontend, the model learns robust representations from raw signals and achieves state-of-the-art performance - including over 95% cross-subject accuracy for epilepsy classification.
A Model with Real Impact
MANAS-1 achieves a remarkable 95% accuracy in distinguishing between epileptic and PNES patients, demonstrating its potential for revolutionizing how AI is used in healthcare
Translating the Language of the Human Brain
MAE-Based Brain Foundation Model
Large-scale MAE pretraining on diverse Western and Eastern EEG datasets. State-of-the-art performance across benchmarks with 95%+ cross-subject epilepsy classification accuracy.
Health & Diagnostics
Advanced neurological screening, disease detection, and clinical decision support powered by foundation-scale EEG models.
Lifestyle & Wellness
Consumer-grade neurotechnology applications for everyday brain insights and optimization.
Bionics & Advanced Robotics
Neural control systems enabling seamless interaction between humans and machines.
Neuro-Symbolic AI
Bridging deep learning with symbolic reasoning for explainable and trustworthy AI.
Neuroscience Research & Drug Development
Large-scale EEG representation learning to accelerate scientific discovery and therapeutic innovation.
