Methods for causal analysis of longitudinal data

医学和健康研究中的一些最基本的问题是：“某种特定的治疗方法是否能使病人活得更长？“或“某种特定的治疗是否减少了疾病发作和个人经历的次数？"。在某些情况下，这些问题可以通过使用随机试验来解决，但这种试验非常昂贵，可能需要多年才能完成。另一种方法是尝试使用所谓的观测数据来回答这些问题。即，已完成研究、登记研究或数据库中的可用数据，其中记录了一组受试者的治疗、结局、患者特征。 这项研究计划的重点是回答这些类型的问题的设置中，可用的数据已被反复测量随着时间的推移。我们的目标是创建一个框架，用于分析数据时，感兴趣的结果是一个经常性的事件过程和个人的个人特征和治疗历史随着时间的推移而演变。我和我的团队还将考虑这样的设置，即关注的结果是生存时间，并且复发事件过程形成了关注的暴露或治疗。解释非随机化治疗和变量之间的复杂关系需要先进的统计方法，我们目前没有足够的方法来处理所有可能的设置。 该方法将由我的团队开发，不仅将促进统计领域内的科学理解，而且可以应用于任何纵向数据和经常性事件发生的实际问题。这可以从与健康相关的应用（疾病的反复发作和/或其与总体生存的关联），精算过程（对保险单的重复索赔）到工业应用（影响机器耐用性或寿命的经常性维修需求）。观测数据的使用令人兴奋，因为它提供了从现有数据中受益并快速获得结果的机会。两名博士生和五名硕士生将在这个研究项目下进行培训，以满足加拿大对训练有素的统计学家和数据科学家日益增长的需求。

Some of the most fundamental questions in medical and health research are: "Does a particular treatment make sick patients live longer?" or “Does a particular treatment reduce the number of episodes of sickness and individual experiences?”. In some settings these questions can be addressed through the use of a randomized trial however such trials are very expensive and may take many years to complete. An alternative is to try and answer these questions using what is known as observational data. That is, data available from completed studies, registries or databases in which the treatments, outcomes, patient characteristics have been recorded for a group of subjects. This research program is focused on answering these types of questions in settings in which the available data has been repeatedly measured over time. The objective is to create a framework for the analysis of data when the outcome of interest is a recurrent event process and an individual's personal characteristics and treatment histories evolve over time. My team and I will also consider settings in which the outcome of interest is survival time and a recurrent event process forms the exposure or treatment of interest. Accounting for the nonrandomized treatment and complex relationship between the variables requires advanced statistical methodology and we do not currently have adequate methods to handle all possible settings. The methodology to be developed by my team and will not only advance scientific understanding within the field of statistics but can be applied to practical problems anywhere longitudinal data and recurrent events occur. This can range from health related applications (recurrent bouts of disease and/or their association with overall survival), actuarial processes (repeated claims on insurance policies) to industrial applications (recurrent need for repairs affecting the durability or lifetime of a machine). The use of observational data is exciting since it offers the opportunity to benefit from existing data and quickly obtain results. Two doctoral students and five master's students will be trained under this research program helping to meet the growing need for highly trained statisticians and data scientists in Canada.