Redd Christian B, Silvera-Tawil David, Hopp Dagney, Zandberg Daniel, Martiniuk Alexandra, Dietrich Colin, Karunanithi Mohan K
Annu Int Conf IEEE Eng Med Biol Soc. 2020 Jul;2020:4273-4277. doi: 10.1109/EMBC44109.2020.9176506.
Children, particularly those with atypical or delayed development, have a reduced ability to self-regulate their emotions and behaviour. After a number of anxiety or stress provoking events, this reduced regulatory ability can result in a meltdown. Extrinsic signals of an impending meltdown are often recognised and acted on by clinicians or parents. These external indications are also accompanied by internal physiological changes, such as increase in heart rate, skin electrodermal activity, and skin temperature. These physiological signals may be used to predict impending meltdown events and facilitate earlier and effective carer intervention, especially in complex management cases. We present a preliminary study using a wearable sensor system for continuous monitoring of physiological signals to measure and predict emotional changes in school-aged children. Our models are able to correctly classify the behavioural state of a child with 68% mean global model accuracy and up to 85% for person-dependent models. Prediction of emotion and identification of impending meltdowns will potentially assist parents, carers, teachers and clinicians to manage stress and problem behaviours before they escalate, and support self-management strategies throughout the variety of normal daily life.
儿童，尤其是那些发育异常或发育迟缓的儿童，自我调节情绪和行为的能力较弱。在经历多次引发焦虑或压力的事件后，这种减弱的调节能力可能会导致情绪失控。即将情绪失控的外在信号通常能被临床医生或家长识别并采取应对措施。这些外部迹象还伴随着内部生理变化，如心率加快、皮肤电活动增加和皮肤温度升高。这些生理信号可用于预测即将发生的情绪失控事件，并促进护理人员更早地进行有效干预，尤其是在复杂的管理案例中。我们开展了一项初步研究，使用可穿戴传感器系统持续监测生理信号，以测量和预测学龄儿童的情绪变化。我们的模型能够以平均68%的全局模型准确率正确分类儿童的行为状态，对于依赖个体的模型，准确率高达85%。情绪预测和即将发生的情绪失控识别将有可能帮助家长、护理人员、教师和临床医生在压力和问题行为升级之前进行管理，并在日常生活的各种场景中支持自我管理策略。
