A Posture monitoring algorithm using a 3 axis acceleration sensor

• Published in: Jinko Zoki Vol. 28; no. 1; pp. 78 - 82
• Main Authors: 藤本 裕, Toyoshima Takeshi, 南 慶一郎, Hashimoto Toru
• Format: Journal Article
• Language: Japanese
• Published: 一般社団法人 日本人工臓器学会 1999; JAPANESE SOCIETY FOR ARTIFICIAL ORGANS
• Subjects: Acceleration sensor; Integration; Pacemaker; Posture monitoring
• ISSN: 0300-0818; 1883-6097
• DOI: 10.11392/jsao1972.28.78

A Posture monitoring algorithm using a 3 axis acceleration sensor. Detection of change in the posture of a patient with an implanted pacemaker provides a useful means for diagnosing and treating fainting caused by hemodynamic changes relating to patients' postures. We conducted a study using a 3-axis acceleration sensor, built into a pacemaker. To obtain the variation and detect the change of posture, we twice integrated the absolute values of the vector of the acceleration generated by a movement. But this algorithm easily produces error caused by the integration of even a small noise or offset of the sensor. This resulted in a low detection rate. Therefore we modified this algorithm along more practical lines so as to detect the postural changes. We subtracted the gravity component using a filter rather than constant value. We limited the integration interval to a period of body movement and compensated for kinetic acceleration using centrifugal force from the rotating motion of the body that accompanies posture change. We recorded acceleration during posture changes from sitting to standing and standing to sitting in 13 subjects. The algorithm detected posture changes by the double integration of acceleration signal with a 92% recognition rate from a total of 474 postural changes. This algorithm generated stable detection. ペースメーカ患者の立位-座位の変化が識別出来れば, 循環機能の変化が誘因となる失神発作等の有効な治療手段となりうる. 我々は, 体幹部の3次元加速度ベクトルの絶対値を2回積分して得られる変位から, 座位から立位, またはその逆の姿勢変化の識別を試みてきた. しかし, 連続積分法では, 加速度センサーのドリフト, 体動により発生する遠心力等の影響により誤差が発生し, 実用性が低かった. 本研究では, アルゴリズムをより実用的なものとするため, i) 高域通過フィルターによる重力加速度の除去, ii) 姿勢変化中のみの加速度積分, iii) 姿勢変化時の身体の回転運動による遠心力に対する補正, などの改良を試みた. 本アルゴリズムにより, 被験者13人について立位-座位問での姿勢変化時の加速度を記録し, 2回積分により変位を求める方法で, 姿勢変化の判定を行った結果, 姿勢変化総数474回に対して92%の正解率で識別でき, 実用性が高いことが判明した.