We implemented a wearable system for detecting accelerations of the user's head while standing still for the purpose of developing a daily health care application. A 2-axis accelerometer was attached on the top of the user's head to separately detect faint accelerations in both the front to back and right to left directions. The total weight of the headset device is only 195 grams including a 9V NiMH battery. This paper presents the results of several experiments performed with the headset system. First, we observed healthy subjects under normal conditions and recorded typical accelerations in the range of 10-30 milli-Gs, which was sufficient to be detected by the system's sensitivity. We then performed numerical analysis on traces of acceleration patterns for a specific user in different conditions. From these analyses we found that the total length of 2-dimensional acceleration pattern trace and a high frequency spectrum (2Hz-10Hz) of right/left acceleration were related to the physical condition of the user. The wearable headset device can be carried to anywhere the user goes and the diagnosis of the wearer's physical condition only requires that the user stand still for 30 seconds, showing that this system can be used for daily health care monitoring.