Emotion-aware assistive technologies in rehabilitation and healthcare

对帮助照顾加拿大日益增长的残疾人和老龄化人口的技术的需求日益增长。当前技术的一个局限性是，它们无法像人类照顾者那样识别情绪或对情绪做出反应。情感感知将增强辅助技术解释和适应用户需求的能力。例如，对严重恐惧的检测可以提高独居老年人跌倒探测器的准确性；对挫折或参与的意识可以用来优化电脑游戏，以进行虚拟学习和康复；对临床重要情绪的警觉可以促进对慢性病(例如抑郁症)的自我管理。许多研究表明，在实验室条件下，可以通过观察心率和呼吸等多种生理信号来检测情绪。然而，在日常消费产品能够可靠地检测到情绪之前，需要科学和工程的进步。本研究的目标是：(1)开发能够适应不同人的不同生理和情绪反应的情绪检测器；(2)通过结合情绪的生理、上下文(例如一天中的时间)和心理(例如个性特征)指标来提高检测的准确性；以及(3)创建一个原型的情绪感知计算机学习游戏，并比较用户在有情绪意识和没有情绪意识的情况下的表现和游戏的乐趣。为了实现这些目标，我们将收集参与者在参与难度和持续时间不同的电脑游戏时的情绪、生理、心理和背景数据。我们将引发与学习体验相关的积极(参与、享受)和消极(沮丧、焦虑)情绪。我们将开发模型，使用收集的数据对体验到的情绪进行分类。我们将在不同的时间针对不同的人测试我们的模型的准确性。这项研究最终将导致为自闭症儿童开发计算机学习游戏，以改善获得与护理相关的强化治疗的机会并降低成本。我们的研究在工作场所压力管理、智能家居和健康监测方面具有广泛的商业机会。

There is a growing need for technologies to help care for Canada's increasing disabled & aging populations. One limitation of current technologies is that they cannot recognize or respond to emotions like human caregivers do. Emotion-awareness would enhance the ability of assistive technologies to interpret & adapt to user needs. For example, detection of acute fear could improve the accuracy of fall detectors for elders living alone; awareness of frustration or engagement could be used to optimize computer games for virtual learning & rehabilitation; alertness to clinically important emotions could facilitate self-management of chronic conditions (e.g. depression). Many studies have shown that emotions can be detected by observing multiple physiological signals like heart rate & respiration in laboratory conditions. However, scientific & engineering advancements are needed before emotions can be reliably detected by everyday consumer products. The objectives of this research are:(1) To develop emotion detectors that can accommodate diverse physiological & emotional responses of different people; (2) To improve detection accuracy by combining physiological, contextual (e.g. time of day) & psychological (e.g. personality traits) indicators of emotions, & (3) To create a prototype emotion-aware computer learning game and compare user performance and enjoyment of the game with and without emotion-awareness. To achieve these objectives, we will collect emotional, physiological, psychological, & contextual data from participants as they engage in computer games that vary in difficulty & duration. We will elicit positive (engagement, enjoyment) & negative (frustration, anxiety) emotions related to the learning experience. We will develop models to classify the emotions experienced using the collected data. We will test the accuracy of our models for different people at different times. This research will eventually lead to the development of computer learning games for autistic children to improve access to and reduce costs of intensive therapies associated with their care. Our research is widely applicable with commercial opportunities in workplace stress management, smart homes & health monitoring.