The research project aims to develop an intelligent system for decoding motor intention through the processing of surface EEG signals. Compared with other neuroimaging methods, the exclusive use of surface EEG makes the system portable, non invasive and low cost.

The approach investigates neuro inspired architectures based on modular blocks of oscillators to analyze EEG sensorimotor rhythms associated with the motor areas of the human brain involved in the execution or imagination of movement.

In addition to enabling rapid subject specific calibration, the system produces interpretable maps of the most relevant EEG patterns used for motor intention identification. This technology can be integrated into Brain Computer Interface systems, contributing to significant advances in neurological rehabilitation and in the control of assistive devices for patients with motor impairments.

The research develops intelligent and interpretable systems for clinical support, integrating electronic devices, signal analysis, computer vision and artificial intelligence.

A central area focuses on motor rehabilitation in patients with neurological disabilities, through platforms integrated with Functional Electrical Stimulation systems controlled by electromyographic signals. The system uses RGB webcams and YOLOv8 Pose algorithms to track upper limb movements in real time, extracting kinematic parameters useful for patient assessment and for adjusting electrical stimulation.

Preliminary tests show the platform’s ability to identify movement events and support therapy adaptation.

The project develops a multimodal system to support the early identification of Autism Spectrum Disorder risk in children aged 13 to 23 months.

The approach integrates EEG, video and audio signals to objectively analyze neurophysiological and behavioral indicators, such as facial expressions, gaze direction, head and body movements, vocalizations and responses to standardized socio communicative tasks.

Predictive models are validated against the clinical diagnosis at 24 months, based on ADOS 2, with the aim of supporting earlier, scalable screening that is less dependent on clinical observation alone.

The use of Explainable AI aims to make the results more transparent and interpretable, providing clinicians with useful tools to identify early conditions that may require further diagnostic assessment.