基于情感心理、生理数据知识表示及推理的抑郁早期预测研究

With the increasing pressure of the daily life, the mental disorder especially depression continues to rise, which results in huge economic losses and social harm. How to discovery the objective and uniform the index system so as to achieve the effective early prediction of depression is still a great challenge. According to the characteristics of the affective psychophysiological data and the application characteristics of early prediction of depression disorder, this project will do research on the method of knowledge representation and reasoning of the affective psychophysiological information, data processing, and multi-modal data fusion, besides, a prototype system for early prediction of depression will be built. Based on the study of these problems, on the one hand, the unified quantitative index system to evaluate the level of depression disorder is expected to reveal, on the other hand, the early prediction theory, method and technology system will be built based on the knowledge representation and reasoning of affective psychophysiological data. This study will provide not only a new way for the early diagnosis and treatment of mental disorder, but also the theoretical and practical reference for the development of knowledge modeling and reasoning in the mental healthcare field.

随着社会生活压力不断加大，当前心理疾病尤其是抑郁障碍发病率不断升高，由此产生巨大的社会危害和经济损失。如何找到客观、统一的抑郁早期量化评估指标体系，从而实现抑郁早期的有效预测仍是一个巨大的挑战。本项目将根据情感心理、生理信息特点，结合抑郁早期预测的应用特征，研究面向抑郁早期预测应用的情感心理、生理数据知识表示和推理建模方法、数据处理方法、多模态数据融合方法，搭建抑郁早期预测原型系统。这些问题的研究，有望揭示抑郁早期评估的客观、统一的量化指标体系，创建一套完整的基于情感心理、生理数据知识表示和推理的抑郁早期预测理论、方法和技术体系，为实现精神心理疾患早诊早治提供新的思路，也为知识建模和推理在精神健康领域的发展提供理论和实践上的参考依据。

以促进抑郁病患早诊、早治的战略需求为引导，本项目根据情感心理、生理信息特点，结合抑郁预测的应用特征，研究了面向抑郁预测应用的情感心理、生理数据知识表示建模方法、数据处理方法。从抑郁识别、抑郁知识图谱构建、情绪诱发形式化建模到共情对话生成等方面搭建抑郁预测原型系统。本项目具体构建了基于脑电信号的多尺度核三维卷积神经网络模型来有效地识别抑郁人群。构建了一个抑郁障碍知识图谱，并基于该知识图谱实现问答，为病患提供抑郁自助查询。构建了智能体情感诱发模型，模型实现对情感智能体心理状态和情感诱发的建模。构建了一个强化学习模型，该模型结合共情的认知和情感方面，提高了共情对话系统的知觉能力和情感表达能力。项目研究为实现抑郁疾患早诊早治提供新的思路，也为知识建模和推理在精神健康领域的发展提供理论和实践上的参考依据。

内容获取失败，请点击重试