针对特定人群的抑郁症早期识别模型研究

Depression is a kind of disease with high prevalence, heavy social burdens and lack of recognitionleads to high misdiagnosis rate.Screening fits for depression and there are screening tools, while because of limited medical resources and weak screening conditions, it is difficult to achieve a wide range of large-scale screening.Depression screening is to identify depression at early stage through a comprehensive assessment of the specific groups of individuals depressive symptoms, natural conditions, social, psychological, personality and other factors.To meet urgent needs, our groups are planning to establish a depression early recognition model for specific populations (college students, staff) and verify the model..Multi-factor analysis will be used to integrate large sample cases of depression and clinical research data, combined with the application of machine learning technology of computational science. To screendepression related factorsand find depression occurrence regularity, to establish and verify early recognition incremental learning style model for depression..The establishment of this early identification model for depression could be achieved accurately and quickly identify and forecast depression. This makes it possible for us to screen depression among larger populations, so as to promote early treatment and prevention for depression, to relieve personal and social burden.

抑郁症患病率高、社会负担重，而识别率低。抑郁症适合筛查且有筛查工具，但因医疗资源的限制，筛查条件很薄弱，很难实现较大范围较大规模的筛查。筛查抑郁症是通过对特定人群中的个体的抑郁相关症状、自然状况、社会、心理、人格等因素的综合评估来早期识别抑郁症。针对紧迫的现实需求，课题组拟建立针对高校学生、公司职员人群的抑郁症早期识别模型，并进行验证。. 课题组拟利用多例抑郁调研资料和抑郁症临床观察数据，应用机器学习技术，结合医学知识，从以上数据中寻找抑郁相关因素与发病的关系规律，建立增量学习式的抑郁症早期识别模型。. 这一抑郁症早期识别模型的建立，可以实现对抑郁症准确、快速的识别和预测，使得在较大规模的人群中进行抑郁症筛查成为可能，从而促进抑郁症的及早治疗和预防，减轻个人和社会负担。

抑郁症患病率高、社会负担重，而识别率低。抑郁症适合筛查且有筛查工具，但因医疗资源的限制，筛查条件很薄弱，很难实现较大范围较大规模的筛查。而建立抑郁症的自动化早期识别模型,可以实现对抑郁症准确、快速的识别和预测,使得在较大规模的人群中进行抑郁症筛查成为可能,从而促进抑郁症的及早治疗和预防,减轻个人和社会负担。.课题组选择前期面向北京市区企业员工和北京及内蒙古高校学生两个人群的抑郁调研数据进行因素分析，同时关于抑郁发病相关知识和经验开展对精神科专家的访谈，筛选抑郁发病相关因素。入选的抑郁症发病相关因素涉及人群中个体的抑郁相关症状、自然状况、社会、心理、人格等多个方面。并结合机器学习的方法，寻找相关因素与发病的关系规律，建立了针对这两个人群的抑郁症早期识别模型，并开发和测试了相应软件。