Hemispheric emerges with $52 million and a six-billion-parameter AI model of the brain
Descartes, trained on 250,000 hours of brain data, analyzes 15 minutes of dry-EEG recordings, starting with PTSD diagnosis.
Hemispheric, co-founded by neuroscientist Hagai Lalazar and Gidi Littwin, a co-inventor of Apple’s Face ID, has come out of six years in stealth with $52 million. Its Descartes model is a six-billion-parameter AI system trained on more than 250,000 hours of brain activity. Patients wear a lightweight dry EEG headset for about 15 minutes while using an app. The first product targets PTSD diagnosis, followed by brain injury, depression, anxiety, schizophrenia and Alzheimer’s. The company aims for availability in 2027, subject to FDA review.
“Foundation models” for brain data are an emerging trend, applying the large-model approach behind chatbots to EEG.
Learn more: Neural decoding · EEG