登录 EN

添加临时用户

基于认知行为疗法的情绪调节严肃游戏设计与实现

CBT-based Affective Game as a Digital Mental Health Intervention

作者:蔡璟荷
  • 学号
    2020******
  • 学位
    硕士
  • 电子邮箱
    asa******com
  • 答辩日期
    2023.05.22
  • 导师
    贾珈
  • 学科名
    计算机科学与技术
  • 页码
    79
  • 保密级别
    公开
  • 培养单位
    024 计算机系
  • 中文关键词
    大学生群体,数字心理健康干预,认知行为疗法,严肃游戏,情绪调节
  • 英文关键词
    College student, Digital mental health intervention, Cognitive behavioral therapy, Serious game, Emotion regulation

摘要

情绪调节是影响心理健康水平和幸福的重要心理能力。日趋显著的大学生心理健康问题表明了该群体情绪调节能力的缺失。认知行为疗法(CBT)因其循证基础、结构清晰、短程高效等特点,已成为化解情绪困扰和认知行为问题等心理障碍的一线治疗方法。然而,治疗资源的匮乏、心理健康素养水平的参差不齐、心理健康服务的隐私性不足等因素限制了大学生获得心理健康干预的机会。在这种情况下,人机交互领域涌现了许多数字心理健康干预方法。严肃游戏(应用于严肃目的的计算机游戏)作为一种极具创新和潜力的数字干预类型,已被应用于不同的心理健康领域,越来越多的研究确定了其临床方面的价值。但由于缺乏统一的设计指导方针,现有的基于CBT的严肃游戏在游戏类型和功能上存在显著的差异, 基于游戏和非游戏干预措施的实证比较也十分有限,导致严肃游戏的设计者难以提取有效的游戏化策略。鉴于大学生群体情绪问题的普遍性和目前CBT严肃游戏研究的局限性,本文以认知行为心理学理论为基础,研究如何通过严肃游戏改善大学生的情绪调节能力。本研究的主要贡献如下: (1)理论性贡献:提出了三视角(游戏视角、技术视角、治疗视角)认知行为游戏设计模型。该模型以玩家体验为核心要素,源自于对十余种游戏设计模型的比较研究,并且通过设计实践,探讨了一个完整有效的CBT严肃游戏的设计框架,为后续基于CBT的严肃游戏设计提供了结构化支持。(2)应用性贡献:针对大学生群体,设计并开发了基于情感和认知改变的互动叙事游戏CatHill,该游戏允许玩家使用呼吸和语音输入以增强玩家的游戏体验,帮助玩家与游戏系统建立良好的治疗联盟。CatHill相比非游戏策略的计算机辅助CBT系统,消极情绪和认知重评等主要衡量指标与自动思维感知等次要衡量指标均有显著提升。(3)技术性贡献: 构建了一种适配游戏场景、以玩家为中心的多模态情感计算方法,该方法在游戏过程中同时收集玩家的生理情绪和认知情绪以获得融合的情感信号,从而更准确地对玩家的情感结果进行识别和分类用于调整游戏的具体特征。本研究分别在目前最大的半监督中文多模态情感数据集和研究中的游戏场景下进行了综合实验,验证了该方法的有效性。

Emotion Regulation (ER) plays a crucial role in one’s health status and well-being. The increasingly significant mental health problems of college students indicate the lack of ER ability of this group. Given its evidenced base, clear structure, and effectiveness, Cognitive Behavioral Therapy (CBT) has been recommended as the first line of treatment for youngsters who suffer from anxiety and depression. However, several factors, including a lack of therapeutic resources, a shortage of mental health literacy, privacy concerns, mental illness stigma, etc., limit college students from accessing traditional psychological aid. In such cases, numerous digital mental health interventions (DMHIs) emerge in the HCI field. Serious games (computer-based games designed for serious purposes) as one type of DMHI are innovative and promising, having been applied to various health interventions. More and more studies have identified its values. But due to the absence of uniform design guidelines, the difference in game genres and features that appeal to and motivate players is significant among existing CBT serious games. Besides, there are limited empirical comparisons between gamified strategies and non-gaming ones.Focusing on the current psychologic problem of college students and the limitation of previous studies in CBT serious games, This research is based on CBT to study how to improve ER ability of university students. The staple contributions of this research are as follows:1. Theoretical contribution: Proposed the GTT (Game perspective, Technical perspective, Therapeutic perspective) serious game design model. This model emphasizes the player experience, derived from a comparative study of twelve game design models, which can be the guidelines for future studies in CBT-based serious games.2. Application contribution: Designed and developed an emotion-based interactive storytelling game named CatHill leveraging CBT to enhance college students‘ mental health. The game allows players to use voice control and breath control, helping establish intimacy between players and their characters. It also facilitates players to set up a therapeutic alliance with the DMHI application, which is the “precondition” for improving clinical effectiveness.3. Technical contribution: Constructed a multi-model affective computing method that extracts the fusion emotion signals from the audio, the corresponding texts, and the consecutive frames of players‘ faces. This approach is player-centered and adapts to gaming scenarios, accurately recognizing and classifying players’ emotional status to adjust the game’s specific features. The effectiveness of the method is verified by conducting comprehensive experiments on the largest semi-supervised Chinese MSA (multimodal sentiment analysis) dataset available and the game scenarios.