报告时间：2026年4月14日（周二）下午 15:00-16:00

报告地点：西汉姆联天赐庄校区（本部）博远楼407

报告人：凌成秀 教授，西交利物浦大学

报告摘要：

Stroke-like episodes (SLEs) are debilitating and recurrent neurological events in mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes (MELAS) syndrome, yet predicting their recurrence remains challenging. Existing prognostic models primarily focus on first-event risk and underutilize longitudinal neuroimaging information collected during routine follow-up. We developed a dynamic framework to forecast recurrent SLEs using longitudinal imaging biomarkers. A total of 83 patients with genetically confirmed mitochondrial disease who experienced recurrent SLEs were analyzed, yielding 144 gap-time intervals between consecutive events. Longitudinal trajectories of 13 MRI-derived biomarkers reflecting brain structure, lesion burden, functional connectivity, and cerebral perfusion were summarized into low-dimensional representations using multivariate functional principal component analysis (MFPC) and incorporated into gap-time Cox models accounting for within-patient dependence. Model performance was evaluated using repeated 10-fold cross-validation with discrimination and calibration metrics. Incorporating longitudinal imaging trajectories substantially improved predictive performance compared with clinical and static imaging models, with approximately 10% improvement in C-index and 12% improvement in integrated AUC. Dynamic predictions enabled individualized recurrence risk estimates that updated as new imaging data accumulated. These findings demonstrate that longitudinal neuroimaging supports personalized, time-adaptive prediction of recurrent SLEs and may facilitate clinical monitoring and trial design in MELAS syndrome.

报告人简介：

凌成秀，博士（洛桑大学保险精算，2014），西浦慧湖药学院高级副教授，博导，江苏省应用统计协会理事（2021-2025），苏州现场统计副理事长（2023-），生物统计硕士专业课程负责人（2025-），江苏省中青年学术带头人（2022-2025），苏州工业园区管委会国际学科领军人才，获西浦慧湖药学院教学创新奖（2025年），优秀博士生导师（2024年），主要研究贝叶斯时空模型及其在极端环境与传染病关联、纵向-生存数据的因果中介分析、脓毒症生物标志物发现等。完成国家自然科学青年基金，中国博士后基金等，发表40+SCI收录文章于Statistics Sinica, Environmental and Ecology Statistics, J. Environmental Management等。

邀请人：唐煜