社区严重精神障碍患者暴力行为复杂因子动态影响机制的统计建模研究

The prevention and control of violence risk in patients with severe mental disorders is a major public health and social security problem that needs to be addressed. However, the mechanism of violence behavior in patients with severe mental disorders, especially the direct/indirect effects of dynamic risk factors which can be intervened on violence behavior are still unclear. Existing researches at home and abroad have not focused on key issues such as multi-exposure and multi-mediation, complex confounding and interaction, temporal and spatial variability of high-dimensional longitudinal data. Based on the data characteristics of multi-source and multi-level structural information of the mega-cohorts and on-site surveys derived from the “National Data System of Severe Mental Illness”, this project plans to (1) develop an advanced approach, Bayesian dynamic multi-level mediation model, and explore its construction strategies and several pivotal technical problems; (2) incorporate social environment (space dimension) into the model based on the included time dimension to investigate the temporal and spatial variability and spatio-temporal interactions by including a random-effects meta-regression model; (3) conduct an in-depth study of the distribution characteristics, changes, and especially the path and mechanism of dynamic risk factors of violence behaviors in patients with severe mental disorders in community. The highlights of this project lie in both its crucial academic and application value in that it will not only act as an essential methodological reference for similar studies, but also provide scientific evidence for targeted prevention and control of violence behaviors in patients with severe mental disorders.

严重精神障碍(简称重精)患者的暴力风险防控是一个亟待应对的重大公共卫生和社会安全问题，但重精患者暴力行为的发生机制，尤其是可干预的动态风险因子对暴力发生的直接和/或间接效应及路径尚不明确。国内外现有研究尚未涉及高维纵向数据的多暴露、多中介、复杂混杂、时间和空间变异及交互等关键问题。本项目拟以“国家重性精神疾病基本数据系统”大型队列资料及补充随访调查等多源多层次结构数据为基础，发展贝叶斯动态多水平中介模型这一先进分析技术，探讨其构建策略及建模的若干关键技术问题；并首创性地将其从时间维度拓展至社会环境(空间)维度，通过引入随机效应meta回归模型，探讨其时间、空间变异及时空交互效应，同时从时空两个维度深入研究重精患者暴力行为的分布特征、变化规律尤其是动态风险因子的作用路径及机制。研究成果将为类似研究提供方法学借鉴，为重精患者暴力行为有针对性防控提供科学证据，具有重要的学术和应用价值。

严重精神障碍(简称重精)患者的暴力风险防控是一个亟待应对的重大公共卫生和社会安全问题，但重精患者暴力行为的发生机制，尤其是可干预的动态风险因子对暴力发生的直接和/或中介效应及路径尚不明确。国内外现有研究尚未考虑高维纵向数据的多暴露多中介、复杂混杂交互、时间和空间变异等关键问题。本项目以“国家重性精神疾病基本数据系统”大型队列资料及现场调查数据等多源多层次结构信息为基础，发展贝叶斯动态多水平中介模型这一先进数据分析技术，探讨其构建策略和建模的若干关键技术问题，同时从时空两个维度深入研究我国社区重精患者暴力行为的分布特征、变化规律尤其是动态风险因子的作用路径及机制。本项目已完成的具体研究包括：重精患者暴力相关行为的流行病学现况及特征、重精患者暴力风险评估与预测、重精患者暴力行为复杂因子动态影响机制研究、重精患者寿命相关问题研究以及重精患者生活质量等相关问题研究共计5个方面。研究成果可为类似研究提供重要方法学借鉴，具有较重要的学术价值，同时为重精患者暴力行为有针对性防控提供科学证据，具有重要应用价值。