使用支持向量机分类器从可穿戴传感器数据中识别香烟烟雾吸入情况。
Identification of cigarette smoke inhalations from wearable sensor data using a Support Vector Machine classifier.
作者信息
Lopez-Meyer Paulo, Tiffany Stephen, Sazonov Edward
机构信息
Department of Electrical and computer Engineering, The University of Alabama, Tuscaloosa, AL 35487, USA.
出版信息
Annu Int Conf IEEE Eng Med Biol Soc. 2012;2012:4050-3. doi: 10.1109/EMBC.2012.6346856.
This study presents a subject-independent model for detection of smoke inhalations from wearable sensors capturing characteristic hand-to-mouth gestures and changes in breathing patterns during cigarette smoking. Wearable sensors were used to detect the proximity of the hand to the mouth and to acquire the respiratory patterns. The waveforms of sensor signals were used as features to build a Support Vector Machine classification model. Across a data set of 20 enrolled participants, precision of correct identification of smoke inhalations was found to be >87%, and a resulting recall >80%. These results suggest that it is possible to analyze smoking behavior by means of a wearable and non-invasive sensor system.
本研究提出了一种独立于个体的模型，用于通过可穿戴传感器检测烟雾吸入情况，该传感器可捕捉吸烟过程中手部到嘴部的特征性动作以及呼吸模式的变化。使用可穿戴传感器来检测手与嘴的接近程度并获取呼吸模式。传感器信号的波形被用作特征来构建支持向量机分类模型。在20名登记参与者的数据集上，发现正确识别烟雾吸入的精度>87%，召回率>80%。这些结果表明，通过可穿戴且非侵入性的传感器系统来分析吸烟行为是可行的。
Zotero 插件