@techreport{oai:ipsj.ixsq.nii.ac.jp:00233197, author = {花当, 祐郁 and 森島, 信 and 浦島, 智 and 峰松, 宏 and 山本, 隆志 and 鳥山, 朋二 and Tasuku, Hanato and Shin, Morishima and Akira, Urashima and Hiroshi, Minematsu and Takashi, Yamamoto and Tomoji, Toriyama}, issue = {25}, month = {Mar}, note = {寝たきり状態にある高齢者の異常を発見することが重要な課題である.しかし,介護士が状態を常に見守ることは困難である.そこで体表面温度を使用した異常状態検知手法を提案する.正常状態では体表面温度と周辺環境の組み合わせパターンが複数存在することを仮定し,そのパターンからの逸脱を異常とする.絶食後の安静状態の体表面温度の変化を正常とし,刺激物やアルコールの摂取時の体表温の変化を異常として提案手法を評価した結果,特徴量を最大値と最小値の差にすることで真陽性率は 78% となった.偽陽性率については 9% となった., Detecting Anomalies in bedridden elderly humans is an important issue. However, it is difficult for caregivers to constantly monitor the condition of the elderly. Therefore, we propose a method of anomaly detection using human body surface temperature. We assume that there are multiple combination patterns of body surface temperature and the surrounding environment in the normal state, and consider any deviation from these patterns as an anomaly. To evaluate this method, we carried out the experiment to detect anomaly in the change of the body surface temperature during stimulating foods or alcohol intake, assuming one in the resting state were considered normal. By using difference between maximum and minimum value as the feature value, the proposed method detected anomaly with a true positive rate of 78%. The false positive rate was 9%}, title = {体表面温度による高齢者の異常状態検知手法}, year = {2024} }