The face can register change before a person feels it.

A symptom that is easy to miss

Reduced facial expressivity is a recognised feature of Parkinson’s, sometimes called hypomimia, and changes in spontaneous expression also show up in dementia. These are exactly the kind of slow, subtle shifts that a busy clinician, or a family member who sees someone every day, can struggle to notice until they are pronounced. By then the window for early action has often narrowed.

What facial-analysis research shows

A 2025 systematic review and meta-analysis of facial-expression deep-learning models reported that they could distinguish people with neurological disorders, including dementia, from healthy controls across pooled studies (Yoonesi et al., BioMedical Engineering OnLine 2025). Separate work using emotion-elicitation video found facial analysis could support early detection and differential diagnosis of cognitive impairment. The appeal is practical: it needs only a camera.

Where this fits in Deep Medicine’s approach

Facial expression is one of the five streams Deep Medicine captures during a single smartphone session, alongside voice, gaze, balance and cognition. It is intended to add a signal that many tools ignore, and is designed to be read in combination with the others rather than on its own. The platform also stores no image as an identifying record of a face.