Multimodal datasets and wearable devices target translation

A 50-participant synchronous EEG-fNIRS dataset reported roughly 10% multimodal improvement over the best unimodal model, while a $1.015 million NIH R01 targets integrated EEG-fNIRS-EMG and AI motor decoding. A new offline fNIRS-guided reinforcement-learning study improved learning in simulation, but transfer to people and physical robots is untested. Xinnao’s wireless dry-electrode EEG recorder has Class II registration, while its separate BCI intervention system has only a classification determination; neither establishes BCI efficacy.