With furrowed brows and pursed lips, the patient's face lights up. Through a series of clicks on his smartphone, he's found the secondhand jacket he was looking for. With his consent, each of his facial expressions is analyzed by artificial intelligence (AI) on the device to track its degree of fluctuation. Because in mood disorders such as depression or bipolar disorder, abnormally high or low emotional signatures can indicate a relapse.
Identifying them and perhaps even anticipating them between medical appointments would be " very useful", comments to Science and Future Professor Fabien Vinckier, psychiatrist at GHU Paris, Professor at Paris Cité University and a researcher at the Brain Institute. It is this early detection of abnormal emotional variance through passive monitoring by their AI that the founders of the startup Emobot, which originated from research at CentraleSupélec, hope to demonstrate.
English 
French