According to the clinical needs of the hospital＇s old adult wards, we develop an AI edge computing-based two-stage early bed-exit warning and bedside fall notification system for aged patients with a high risk of falls. The proposed system can effectively help the hospital to reduce the incidence of bed fall and increase the notification rate of bed fall. Moreover, the rate of injury from falling bed-exit can be reduced.Our EyeWall only needs to hang on the wall behind the bed at about 2~2.5 meters. Moreover, AI-based human torso recognition technology is adopted. Then the human skeleton and joint point movement ratio are processed for judging related continuous actions (lying in bed→getting up and sitting up→sitting on the edge of the bed→bed-exit→bedside fall). Furthermore, the bed-exit warning and bedside fall event will also be sent to the registered professional nurse mobile phone or nursing station through the Internet.This system is mainly for the safety of elderly patients hospitalized in medical institutions. It introduces relevant AI and ICT technologies to construct a scientific and technological care environment for providing appropriate care services immediately.