Avoiding C-hacking when evaluating survival distribution predictions with discrimination measures
Imperial College London(伦敦帝国理工学院);Technische Universität Kaiserslautern(凯泽斯劳滕工业大学);LMU Munich(慕尼黑大学);Deutsches Forschungszentrum für Künstliche Intelligenz(德国人工智能研究中心);University of Warwick(华威大学)
Raphael Sonabend, Andreas Bender, Sebastian Vollmer
In this paper we consider how to evaluate survival distribution predictions with measures of discrimination. This is a non-trivial problem as discrimination measures are the most commonly used in survival analysis and yet there is no clear method to derive a risk prediction from a distribution prediction. We survey methods proposed in literature and software and consider their respective advantages and disadvantages. Whilst distributions are frequently evaluated by discrimination measures, we find that the method for doing so is rarely described in the literature and often leads to unfair comparisons. We find that the most robust method of reducing a distribution to a risk is to sum over the predicted cumulative hazard. We recommend that machine learning survival analysis software implements clear transformations between distribution and risk predictions in order to allow more transparent and accessible model evaluation. The code used in the final experiment is available at https://github.com/RaphaelS1/distribution_discrimination.
A Large-Scale Neutral Comparison Study of Survival Models on Low-Dimensional Data
Department of Statistics, LMU Munich(慕尼黑大学统计学系);Munich Center for Machine Learning(慕尼黑机器学习中心);Leibniz Institute for Prevention Research and Epidemiology – BIPS(预防研究与流行病学研究所 – BIPS);Faculty of Mathematics & Computer Science, University of Bremen(不莱梅大学数学与计算机科学学院);Department of Cancer Genetics, Institute for Cancer Research, Oslo University Hospital(癌症遗传学部,癌症研究所,奥斯陆大学医院);Department of Biostatistics, Oslo Centre for Biostatistics and Epidemiology (OCBE), University of Oslo (UiO)(生物统计学部,奥斯陆生物统计与流行病学中心(OCBE),奥斯陆大学);OSPO Now, London, UK(OSPO Now,伦敦,英国)
Predicting cardiovascular risk from national administrative databases using a combined survival analysis and deep learning approach
University of New South Wales(新南威尔士大学);University of Auckland(奥克兰大学);National Drug and Alcohol Research Centre(国家药物与酒精研究中心)
目标。本研究比较了生存分析模型的深度学习扩展与传统Cox比例风险(CPH)模型在国家卫生管理数据集中推导心血管疾病(CVD)风险预测方程的性能。方法.使用多个管理数据集的个人联系,我们构建了一个年龄在30-74岁之间的所有新西兰居民的队列,这些居民在2012年期间与公共资助的卫生服务进行了互动,并在五年的随访中确定了心血管疾病的住院和死亡。在排除了先前患有CVD或心力衰竭的人之后,开发了性别特异性深度学习和CPH模型,以估计五年内致命或非致命CVD事件的风险。在整个研究人群和特定风险组的模型之间比较了解释的事件发生时间、校准和区分的比例。调查结果。首次CVD事件发生在2,164,872人中的61,927人中。在诊断和手术中,深度学习模型与女性吸烟(2.04,95%CI:1.99-2.10)和男性慢性阻塞性肺疾病伴急性下呼吸道感染(1.56,95%CI:1.50-1.62)相关的“局部”风险比最大。其他已确定的预测因子(例如高血压、胸痛、糖尿病)与当前关于CVD风险预测因子的知识一致。深度学习模型在解释事件发生时间的比例(Royston和Sauerbrei的R平方:女性为0.468 vs. 0.425,男性为0.383 vs. 0.348)、校准和辨别力(所有p<0.0001)方面显著优于CPH模型。解释。生存分析模型的深度学习扩展可以应用于大型卫生管理数据库,以导出比传统CPH模型…
Flexible Group Fairness Metrics for Survival Analysis
Technische Universität Kaiserslautern(凯泽斯劳滕工业大学);University of Cambridge(剑桥大学);Imperial College London(帝国理工学院);Ludwig-Maximilians-Universität München(慕尼黑大学);J.P. Morgan AI Research(摩根大通人工智能研究院);Chalmers Technical University(查尔姆斯理工大学);University of Gothenburg(哥德堡大学);Leibniz Institute for Prevention Research and Epidemiology - BIPS GmbH(莱布尼茨预防研究与流行病学研究所-BIPS有限公司);Birla Institute of Technology and Science(贝拉理工学院)
A reproducible and extensible framework for benchmarking competing risks survival models
Cancer Research UK Scotland Centre(英国癌症研究苏格兰中心);Institute of Genetics and Cancer(遗传与癌症研究所);University of Edinburgh(爱丁堡大学);University of Augsburg(奥格斯堡大学)
Communication-efficient distributed hazard difference estimation for heterogeneous multi-site survival data
College of Mathematics, Taiyuan University of Technology, China(太原科技大学数学学院);Centre for Quantitative Medicine, Duke-NUS Medical School, Singapore(新加坡国立大学医学院定量医学中心);Systems Research, Duke-NUS Medical School, Singapore(新加坡国立大学医学院系统研究)
When Are Scoring Rules Proper? Bridging Theory and Practice in Survival Model Evaluation
Department of Cancer Genetics, Institute for Cancer Research, Oslo University Hospital(癌症遗传学系,癌症研究所,奥斯陆大学医院);Department of Biostatistics, Faculty of Medicine, Oslo Centre for Biostatistics and Epidemiology (OCBE), University of Oslo, Norway(生物统计学系,医学学院,奥斯陆生物统计与流行病学中心(OCBE),奥斯陆大学,挪威);OSPO Now, London, UK(伦敦,英国OSPO Now);Department of Mathematics, Faculty of Mathematics and Natural Sciences, University of Oslo, Norway(数学系,数学与自然科学学院,奥斯陆大学,挪威);Department of Statistics, LMU Munich, Germany(统计学系,慕尼黑大学);Munich Center for Machine Learning (MCML), LMU Munich, Germany(慕尼黑机器学习中心(MCML),慕尼黑大学,德国);Statistical Consulting Unit (StaBLab), LMU Munich, Germany(统计咨询单位(StaBLab),慕尼黑大学,德国);Leibniz Institute for Prevention Research and Epidemiology - BIPS, Bremen, Germany(预防研究与流行病学研究所—BIPS,不莱梅,德国);Faculty of Mathematics and Computer Science, University of Bremen, Germany(数学与计算机科学学院,不莱梅大学,德国)
Value-Monotonicity Matters: A Concordance Loss for Deep Survival Prediction
Department of Radiation Oncology, University of Colorado School of Medicine(科罗拉多大学医学院放射肿瘤学系);Department of Radiation Oncology, University of Texas Southwestern Medical Center(德克萨斯大学西南医学中心放射肿瘤学系)