CAREER: Mobile Sensor-Based Adaptive Emotion Prediction and Feedback Delivery

可穿戴、移动技术和物联网的最新进展使我们能够收集有关移动模式、睡眠、中枢和外周神经系统活动以及社交互动的即时生理、行为、社交和环境数据，所有这些都不会扰乱日常生活。这些数据表明，我们有可能彻底改变我们诊断、改善和预防健康疾病的方式。该项目设计、实现和评估个性化的自适应算法，以使用多模式传感器数据来检测和预测情绪状态，并向用户提供反馈以改进心理状态的管理。情绪检测和反馈算法适应人类生理、行为、背景和偏好的变化。这项研究将带来情感辅助技术，这些技术可以提高人类的表现、健康和福祉，从而提高生活质量。该项目将提供一个平台，用于整合移动传感器数据，并向对广泛人群有价值的对象提供反馈，以实现个性化医疗。该项目将对人类数据和人工智能技术的开发、评估和使用中的伦理问题提出见解。研究活动将培养研究生，融入数据科学的课堂课程，并为本科生和高中生提供研究机会和暑期实习机会。该项目将开发动态情感建模和反馈系统，以利用多模式人类数据来自动识别和预测人类情感。该系统将提供安全和个性化的反馈信息，以帮助管理情绪状态。这一目标将通过解决三个基本研究挑战来实现：(1)多模式、多时间尺度的生理和行为模式的解释和分析；(2)用于有效情绪检测和预测的自适应情绪标签采样；(3)安全和可持续的自动个性化反馈。由此产生的情绪检测和反馈系统将被集成，以帮助健康状况不佳的用户管理情绪。这些技术的性能、可用性、安全性和有效性将通过人体研究进行评估。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Recent advances in wearable, mobile technologies and the Internet of Things enable us to collect moment-to-moment physiological, behavioral, social, and environmental data on mobility pattens, sleep, central and peripheral nervous system activity, and social interactions, all without disrupting daily routines. Such data show a potential for revolutionizing how we diagnose, ameliorate, and prevent health disorders. This project designs, implements, and evaluates personalized, adaptive algorithms to detect and predict emotional states using multimodal sensor data, and to provide feedback to users to improve management of mental state. The emotion detection and feedback algorithms adapt to changing human physiology, behavior, context, and preferences. This research will result in emotion assistive technologies that enhance human performance, health, and wellbeing, thus improving quality of life. The project will provide a platform for integrating mobile sensor data and providing feedback to subjects valuable to a broad range of populations for personalized medicine. The project will yield insights into ethical issues in the development, evaluation, and use of human data and artificial intelligence technology. The research activities will train graduate students, be integrated into classroom curricula in data science, and provide research opportunitites and summer internships for undergraduate and high school students. This project will develop dynamic emotion modeling and feedback systems to harness multimodal human data for automatic human emotion recognition and prediction. The system will provide safe and personalized feedback delivery to help manage emotional states. This goal will be achieved by addressing three fundamental research challenges: (1) multi-modal, multi-timescale physiological and behavioral pattern interpretation and analysis; (2) adaptive emotion label sampling for effective emotion detection and prediction; and (3) safe and sustainable automatic personalized feedback delivery. The resulting emotion detection and feedback systems will be integrated to help users in poor health manage emotion. The performance, usability, safety, and effectiveness of the technologies will be evaluated via human subject studies. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.