A non-invasive Neurosity BCI was beta-tested for depression support, and a neurogame was developed to actively influence brain activity. LLMs were tested on a Materna server and locally to support therapists diagnostically. The LLMs were benchmarked and a function for working with medical guidelines was implemented. Physical AI was examined, showing speech-enabled robots can help. Runtime analysis enabled wearable medical data to be integrated into diagnostics, and this data was generated and analyzed for correlations.
Research area(s)
Neuroinformatics, data science, LLM engineering, cognitive science
Technical features
Large language models, EEG (electroencephalography), non-invasive BCI, Python, ollama, OpenWebUI, Docker, Kubernetes, bash, shell, javascript, fine-tuning, RAG, focus score, calm score, alphasymmetry, MI (motor imagery), depression guidelines, RagFlow, Neurosity Python SDK, BrainFlow, LSL (Layer Streaming Lab)
Integration constraints
None except Neurosity Python SDK – Data is utilised by Neurosity company
Targeted customer(s)
Medical field (Psycho Therapy, Neurology), Gaming industry, wearables industry, R&D (science labs)
Conditions for reuse
Individual, separate contract agreements
Contact
Rainer Feinen
Email Rainer.Feinen@materna.group