ISSN 1671-3710
CN 11-4766/R
主办:中国科学院心理研究所
出版:科学出版社

心理科学进展 ›› 2020, Vol. 28 ›› Issue (1): 111-127.doi: 10.3724/SP.J.1042.2020.00111

• 研究前沿 • 上一篇    下一篇

计算精神病学:抑郁症研究和临床应用的新视角

区健新1,2, 吴寅1,2, 刘金婷1,2, 李红1,2,3()   

  1. 1深圳大学心理学院
    2 深圳市情绪与社会认知科学重点实验室
    3 深圳市神经科学研究院, 深圳 518060
  • 收稿日期:2019-01-29 出版日期:2020-01-15 发布日期:2019-11-21
  • 通讯作者: 李红 E-mail:lihongszu@szu.edu.cn
  • 基金资助:
    国家自然科学基金(31671150, 31600928);广东省普通高校创新团队建设项目(2015KCXTD009);广东省省级基础研究及应用研究重大项目(2016KZDXM009);深圳市基础研究布局项目(JCYJ20150729104249783, JCYJ20170818110103216, JCYJ20170412164413575);深圳市孔雀计划项目(KQTD2015033016104926)

Computational psychiatry: A new perspective on research and clinical applications in depression

OU Jianxin1,2, WU Yin1,2, LIU Jinting1,2, LI Hong1,2,3()   

  1. 1 School of Psychology, Shenzhen University
    2 Shenzhen Key Laboratory of Affective and Social Cognitive Science
    3 Shenzhen Institute of Neuroscience, Shenzhen 518060, China
  • Received:2019-01-29 Online:2020-01-15 Published:2019-11-21
  • Contact: LI Hong E-mail:lihongszu@szu.edu.cn

摘要:

抑郁症是一种复杂而异质的精神疾病, 给全球带来沉重的疾病负担。尽管基于症状学的诊断方法已被广泛应用于各领域, 但这种方法并不利于病理机制的探讨。另外, 该诊断方法预测效度较低, 导致其难以准确评估和比较各种治疗方案的疗效。计算精神病学方法则能通过理论驱动和数据驱动两种互补的方法解决上述问题, 从而提高对抑郁症的认识、预防和治疗。理论驱动方法基于经验知识或假设, 利用计算建模方法对数据进行多水平分析; 数据驱动方法则基于机器学习算法分析高维数据, 提高抑郁症诊断和预测的准确性, 进而提高治疗的精准度。理论驱动和数据驱动方法的发展与结合, 以及人才和资源的整合, 将会更有效地推进抑郁症的防治。

关键词: 抑郁症, 计算精神病学, 计算模型, 机器学习, 诊断, 治疗

Abstract:

Depression, a complex and heterogeneous mental disorder, leads to great global burdens of disease. Although diagnosis based on nosology is broadly used in several domains, it is still unable to direct the exploration of pathological mechanism of depression. In addition, several treatments developed by this diagnosis have poor outcomes due to its low prediction validity. Computational approaches to psychiatry remedy those limitations and help to improve understanding, prediction and treatment for depression by two complementary approaches: data-driven and theory-driven. Theory-driven approaches apply models to multiple levels of analysis from the prior knowledge or hypojournal of depression. Data-driven approaches, however, adopt machine-learning methods to analyze high-dimensional data to improve the diagnostic and predictive accuracies of depression, and eventually, promote the treatment effects. With the development and combination of these two approaches as well as the integration of resources, it is promising to cure depression and prevent it from occurrence.

Key words: depression, computational psychiatry, computational models, machine learning, diagnosis, treatment

中图分类号: