In recent years, people with swallowing disorders (dysphagia) have increased because of an aging population. More than 5,000 people die of asphyxiation each year after consuming solid foods in Japan. This study was conducted to develop an abnormal detection algorithm for a meal, especially the universal choking sign (hands crossed over the throat). In this study, magnetic position and orientation sensors were attached to three anatomical landmarks on the upper body of subjects and were used to measure human motion during a meal. Support vector machine was used for simulated choking sign detection Experimental results demonstrated that the proposed method detected simulated choking signs during a meal with accuracy of 93.7±4.1 %.