北航经管学院“经济统计论坛”系列讲座
（2024年第4期，总第25期）
北京大学梁宝生副研究员讲座通知
讲座题目：Regression Analysis of Linear Transformation Cure Model in the Presence of Partially Observed Cure Status with Applications to SARS-COV-2 study
讲座时间：2024年6月12日（周三），15:00-16:30
会议地址：新主楼A949
讲座嘉宾：梁宝生 副研究员
讲座嘉宾 简介
梁宝生，北京大学公共卫生学院生物统计系，副研究员。2016年博士毕业于北京师范大学概率论与数理统计专业，2013至2016年博士在读期间先后以联合培养博士生和助研身份分别在美国北卡罗来纳大学教堂山分校生物统计系和美国纽约哥伦比亚大学生物统计系进行访问交流；2018年于香港大学统计及精算系博士后出站，就职于北京大学医学部工作至今。感兴趣的研究领域为生存分析、半参数有效估计以及统计学习算法等，在震后PTSD、帕金森疾病、阿尔兹海默病、肺癌和家族乳腺癌等疾病的临床复杂数据的建模方法等方面开展了系列研究。在Biometrika、Statistica Sinica等国内外期刊发表论文40余篇, 合作完成译著2部，教材2部。 主持国家自然科学基金青年项目1项，主持北京市自然科学基金1项，以骨干成员参加国自然面上项目2项，百度基金1项，首都卫生发展科研专项基金1项。
邀请人：王珊珊 副教授
讲座概要
In conventional mixture cure models, the cure status is completely observable, and the Cox proportional hazards (PH) model is commonly adopted for the latency model. However, in practice the cure status can be partially observed and the PH assumption can be violated. A more flexible and comprehensive mixture cure model with fewer assumptions is needed to tackle these challenges. In this article, we propose a general linear transformation cure model to incorporate the partially observed cure rate, encompassing the PH model and accelerated failure time (AFT) model as special cases. The concordance measure, K index, is employed to assess the discriminatory accuracy of the proposed model under various distributions. Extensive simulations demonstrate that the proposed model outperforms the existing method under several scenarios. Moreover, the proposed model exhibits robustness to the survival time distributions of the uncured individuals. Finally, the proposed method is used to investigate the vaccine efficacy of SARS-CoV-2 in a real word study with 23,467 SARS-CoV-2 cases in Khyber Pakhtunkhwa province, in Pakistan, from March 16, 2020, to November 30, 2021. Our findings emphasize the importance of accurate models and detailed case information for risk identification.
