Machine Learning for Discovery of Patient Journey-Wide Phenotypes and Colorectal Cancer Stratification

结直肠癌是癌症死亡率的第二大原因，全世界每年有88万人死于结直肠癌。这个项目寻求开发新的方法来对结直肠癌患者进行分层，以帮助为临床决策提供信息。例如，虽然部分II期结直肠癌患者从化疗中受益，但要确定哪些特定患者将受益可能是一项挑战[Kannarkatt等人。肿瘤学实践杂志2017年]。尖端信息学技术将被应用于大型数据集，包括大量相关的临床和人口数据，以发现定义整个患者过程中新表型的个人特征的指纹。这些数据驱动的患者表型可能包括影响推动癌症进展的分子过程的因素，例如与生活方式有关的因素。因此，患者表型的发现可能为发展新的表型特异性分子分层方法定义新的队列。在这四年的学习期间，工作将主要在贝尔法斯特女王大学的奥弗顿小组进行，并与英国、威尔士和北爱尔兰的健康数据研究实质性网站有关。学生期间将在爱丁堡的LifeArc诊断开发中心度过六个月，该中心是一个通过ISO13485认证的环境。学生将从LifeArc丰富的诊断开发专业知识中受益，帮助确保预期的新型诊断软件能够胜任潜在的临床应用。

Colorectal cancer is the second highest cause of cancer mortality, associated with >880,000 deaths per annum worldwide. This project seeks to develop novel approaches for stratification of colorectal cancer patients in order to help inform clinical decision-making. For example, while a proportion of stage II colorectal cancer patients benefit from chemotherapy, it can be challenging to identify which specific patients will benefit [Kannarkatt et al. Journal of Oncology Practice 2017]. Cutting-edge informatics techniques will be applied to large datasets, including substantial linked clinical and demographic data, in order to discover fingerprints of individual characteristics that define new phenotypes across the patient journey. These data-driven patient phenotypes may include factors, for example relating to lifestyle, that influence the molecular processes driving cancer progression. Therefore discovery of patient phenotypes may define new cohorts for development of novel phenotype-specific molecular stratification approaches. Work during this four year studentship will be primarily based in the Overton group at Queen's University Belfast and associated with the Health Data Research UK Wales and Northern Ireland substantive site. The studentship includes six months to be spent at the LifeArc Centre for Diagnostics Development in Edinburgh, an ISO13485 certified environment. The student will benefit from LifeArc's considerable diagnostics development expertise, helping to ensure the anticipated novel diagnostic software is competent for potential clinical use.