According to WIRED, Littwin left Apple in 2020 to co-found Hemispheric, a startup training deep learning models on brain data to detect cognitive disorders and disease. The company claims a dataset of 100,000 brain scans and has now closed a $52 million funding round to move toward non-invasive diagnostic tools.
This is the familiar Silicon Valley playbook applied to neuroscience: take sensor expertise, pile on data, promise pattern recognition, then head for medicine. If Hemispheric succeeds, brain diagnostics become cheaper and accessible; if the AI model fails to generalize beyond its training set, patients get expensive false confidence wrapped around a black box.
The Apple pedigree is the headline. The hard part is medicine. FaceID had to recognize a face in milliseconds; a brain model has to sort signal from noise, survive FDA review, and answer the question that matters most: does a system trained on 100,000 scans actually work on the patient sitting in front of you when the cost of being wrong is a missed stroke or early dementia?
Filed to the Technology desk · Jul 15