Identification of Distinct Multimodal Biotypes of PTSD Using Data Driven Approach: A Multisite Big Data Study

使用数据驱动方法识别 PTSD 的独特多模式生物型：多站点大数据研究

• 批准号: 10317107
• 负责人: Xi Zhu
• 金额: $ 17.59万
• 依托单位: NEW YORK STATE PSYCHIATRIC INSTITUTE DBA RESEARCH FOUNDATION FOR MENTAL HYGIENE, INC
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-12-10 至 2024-11-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10317107
• 关键词: Address; Advanced Development; Anxiety; Anxiety Disorders; Behavioral; Big Data; Biological; Biological Markers; Brain; Brain imaging; Categories; Classification; Clinical; Clinical Treatment; Clinical Trials; Cluster Analysis; Complex; DSM-IV; Data; Data Set; Dependence; Development; Diagnosis; Diagnostic; Dimensions; Disease; Ensure; Environment; Fostering; Fright; Functional disorder; Future; Goals; Grant; Heterogeneity; Individual; Knowledge; Machine Learning; Magnetic Resonance Imaging; Major Depressive Disorder; Measures; Mental Depression; Mentored Research Scientist Development Award; Mentorship; Methods; Modeling; Multimodal Imaging; National Institute of Mental Health; Neurobiology; Patients; Pattern; Pattern Recognition; Pharmaceutical Preparations; Post-Traumatic Stress Disorders; Prediction of Response to Therapy; Prevalence; Psychiatry; Psychoses; Psychotherapy; Public Health; Reproducibility; Research; Research Personnel; Research Project Grants; Research Support; Research Training; Rest; Sample Size; Selection for Treatments; Subgroup; Supervision; Symptoms; Techniques; Training; Translational Research; Trauma; Treatment outcome; Work; advanced analytics; anxiety-related disorders; associated symptom; authority; base; clinical heterogeneity; clinical phenotype; comorbidity; data fusion; design; experience; gray matter; individualized medicine; large datasets; machine learning model; memory process; multimodal data; multimodal neuroimaging; multimodality; neurobiological mechanism; patient oriented; predicting response; programs; relating to nervous system; reward processing; skills; supervised learning; therapy development; trauma exposure; treatment comparison; treatment research; treatment response; treatment strategy; unsupervised learning; working group

Posttraumatic stress disorder (PTSD) is a highly prevalent and debilitating disorder. Despite efforts to characterize the pathophysiology of PTSD and its heterogenity, no objective biomarker have been established to aid in diagnosis, and prediction of treatment response. This K01 presents a program for research and training that will support the applicant on a path towards becoming an independent investigator, focused on utilizing a data-driven computational approach and machine learning techniques to identify multimodal neural biomarkers of PTSD (supervised) and multimodal biotypes of PTSD (unsupervised) and explore whether such biotypes could be used to predict response to prolonged exposure (PE), the first line treatment for PTSD. The training plan builds on the candidate’s prior training and experience and capitalizes on a mentorship team and a research environment to foster development of the candidate’s expertise in 1) the neural and behavioral basis of PTSD and anxiety disorders; 2) multimodal data fusion analysis and latent dimension interpretation with data-driven computational approaches and data reproducibility; and 3) patient-oriented translational research in anxiety disorders. This research project will apply both supervised and unsupervised machine learning techniques on multimodal MRI data from the largest existing PTSD dataset (N~3000 from the ENIGMA-PTSD working group). Biotypes identified from this large dataset will then be extended to clinical treatment data. The results of the proposed research will be vital to aid in finding neural biomarkers of PTSD and better predict different treatment outcomes through different biotype targets and will lead to a future R01 grant examining brain-symptoms association across anxiety and trauma-related disorders, and to use the newly identified PTSD biotypes to inform different treatment outcomes in a following R61/33. Together, the research and training experiences and expertise developed through this K01 award will support the applicant’s transition to research independence and ensure the applicant becomes a leading authority in the application of data-driven computational approaches in psychiatry research, and provide the basis for future NIMH grants to explore biotypes from multimodal brain imaging using data-driven computational approaches across anxiety-related disorders.

创伤后应激障碍（PTSD）是一种高度流行和衰弱的疾病。尽管努力 描述PTSD的病理生理学及其异质性，尚未建立客观的生物标志物 以帮助诊断和预测治疗反应。本K 01提出了一个研究和培训计划 这将有助于申请人成为独立调查员，专注于利用 识别多模态神经生物标志物的数据驱动的计算方法和机器学习技术 的PTSD（监督）和多模式生物型的PTSD（无监督），并探讨这些生物型是否可以 用于预测对长期暴露（PE）的反应，这是PTSD的一线治疗。培训计划 建立在候选人之前的培训和经验，并利用导师团队和研究 环境，以促进候选人在1）创伤后应激障碍的神经和行为基础方面的专业知识发展 和焦虑症; 2）多模态数据融合分析和潜在维度解释与数据驱动 计算方法和数据再现性; 3）以患者为导向的焦虑转化研究 紊乱该研究项目将监督和无监督机器学习技术应用于 来自最大的现有PTSD数据集的多模态MRI数据（来自ENIGMA-PTSD工作组的N~3000）。 从这个大型数据集中识别的生物型将扩展到临床治疗数据。的结果 拟议中的研究对于帮助寻找PTSD的神经生物标志物和更好地预测不同的治疗方法至关重要 通过不同的生物型目标的结果，并将导致未来的R 01赠款检查大脑症状 焦虑和创伤相关疾病之间的联系，并使用新发现的PTSD生物型来告知 在随后的R61/33中的不同治疗结局。总之，研究和培训经验， 通过这个K 01奖开发的专业知识将支持申请人向研究独立性的过渡， 确保申请人成为数据驱动计算方法应用的领先权威， 精神病学研究，并为未来NIMH赠款探索多模态大脑的生物型提供基础 使用数据驱动的计算方法对焦虑相关疾病进行成像。