ReLiSyR-Mental Health: Repurposing Living Systematic Reviews to inform drug selection for clinical trials in mental health

我的博士学位将开发一个重新利用生活系统评论，这将策划和分析与心理健康临床试验相关的证据流。除了考虑单一药物干预外，ReLiSyR-Mental Health还将考虑联合药物干预、非药物治疗和治疗方式的组合。我将使用三角测量方法整合临床前数据、临床数据和大规模人类遗传学研究的信息。此外，我将开发贝叶斯排序方法，为后续研究确定干预措施的优先级。在初始阶段，我将开发用于精神病、抑郁和焦虑相关的临床前和临床体内研究的系统在线生活证据摘要（SOLES），打算在ReLiSyR-Mental Health的开发中利用这些平台。系统在线生活证据摘要（SOLES）是一个互动平台，旨在提供与给定研究领域相关的文献的全面概述。SOLES使用有监督的NLP筛选算法来识别与给定研究领域相关的出版物。该方法旨在优化SOLES平台的特殊性，同时保持其开发工作流程的效率。SOLES平台的一个显著特点是，它们拥有已经通过初始自动筛选过程的研究语料库。通过自动化全文检索和采用文本挖掘方法，可以从全文中提取研究文章的标题和摘要（TiAb）文本中不包含的PICO元素，从而比生物医学图书馆中通常可能的更精细地描述可用文献。对时间和研究人员可用性的限制往往阻碍了全面的证据合成，对论文的双重筛选纳入尤其需要耗费大量资源。此外，在主要研究文章的标题和摘要（TiAb）文本的字数限制内包括所有相关的PICO（患者，干预，比较，结果）元素的挑战意味着标准文献检索通常无法捕获所有相关研究。在相关的SOLES平台中进行文献检索，而不是在生物医学图书馆中进行文献检索，可以减少不相关引文的检索，同时提高检索的敏感性。这可以优化系统评价研究选择的效率和完整性，从而提高证据合成的深度和质量。第二个博士目标将是验证这个研究选择方法的方法。

My PhD will develop a Repurposing Living Systematic Review which will curate and analyse evidence streams relevant to clinical trials in mental health. In addition to considering single drug interventions, ReLiSyR-Mental Health will consider combination drug interventions, non-pharmacological treatments, and combinations of treatment modalities. I will use triangulation approaches to integrate information preclinical data, clinical data and large-scale human genetics research. Further, I will develop Bayesian ranking approaches to prioritising interventions for subsequent research. In the initial phase, I will develop Systematic Online Living Evidence Summaries (SOLES) for preclinical and clinical in vivo research related to psychosis, depression, and anxiety, intending to utilise these platforms in ReLiSyR-Mental Health's development. Systematic Online Living Evidence Summaries (SOLES) are interactive platforms which seek to provide a comprehensive overview of the literature relevant to a given area of research. SOLES use supervised NLP screening algorithms to identify publications relevant to a given research area of interest. This methodology seeks to optimise specificity of the SOLES platform whilst maintaining efficiency of their development workflow. A distinguishing feature of SOLES platforms is that they hold a corpus of studies that has already been through an initial automated screening process. By automating full-text retrieval and employing text-mining approaches, PICO elements not included within the Title and Abstract (TiAb) text of a research article can be extracted from the full-text, allowing for a more granular characterisation of the available literature than typically possible in biomedical libraries.Constraints on time and researcher availability often impede comprehensive evidence synthesis, with dual-screening of papers for inclusion being particularly resource intensive. Further, the challenge of including all pertinent PICO (Patient, Intervention, Comparison, Outcome) elements within the word count limitations of Title and Abstract (TiAb) text in primary research articles means that standard literature searches often fail to capture all of the relevant research. Conducting a literature search within the relevant SOLES platform, as opposed to a biomedical library, may reduce the retrieval of irrelevant citations whilst enhancing the search's sensitivity. This may optimise the efficiency and completeness of study selection for systematic reviews, and thus enhance the depth and quality of the evidence synthesis. A secondary PhD aim will be to validate the methodology of this study selection approach.