Machine learning predictors have become essential tools for guiding automated decision making. However, a major misalignment persists: predictive models are typically optimized in terms of standard statistical metrics in isolation from the algorithmic tasks they inform. We highlight this incongruity in the high-stakes domain of organ allocation by demonstrating that any algorithm relying on (even highly accurate) survival predictors optimized for standard metrics -- such as the Concordance index (C-index) -- can yield arbitrarily poor outcomes when used for allocation, failing to guarantee utility better than a uniform random selection. To bridge the gap between survival analysis and policy optimization, we introduce a decision-focused learning approach based on optimizing normalized discounted cumulative gain (NDCG), a mainstay metric in information retrieval. We establish the utility of NDCG in survival analysis by proving that it translates to guarantees on the performance of allocation. Empirically, we propose a bootstrapping approach to optimize the NDCG of existing survival models. Unlike prior work, we also address the challenge of right censorship when evaluating ranking. On historical heart transplant data from the US, our method dramatically boosts the NDCG of baseline models by 50-100%, which translates to tens of thousands of additional life years gained annually when deployed for transplant allocation. We anticipate that our framework will find broader applications in decision making with predictions.
Personalized Survival Predictions for Cardiac Transplantation via Trees of Predictors
University of California, Los Angeles(加州大学洛杉矶分校);Nuffield College, Oxford University(牛津大学纳菲尔德学院);Oxford Man Institute, Oxford University(牛津大学牛津曼研究所);Alan Turing Institute(艾伦·图灵研究所)
Calibration of prediction rules for life-time outcomes using prognostic Cox regression survival models and multiple imputations to account for missing predictor data with cross-validatory assessment
Self-Consistent Equation-guided Neural Networks for Censored Time-to-Event Data
Harvard Pilgrim Health Care Institute(哈佛朝圣者医疗保健研究所);Harvard Medical School(哈佛医学院);Harvard School of Public Health(哈佛公共卫生学院);North Carolina State University(北卡罗来纳州立大学)
Comprehensive Benchmarking of Machine Learning Methods for Risk Prediction Modelling from Large-Scale Survival Data: A UK Biobank Study
King’s College London(伦敦国王学院);British Heart Foundation Centre of Research Excellence(英国心脏基金会卓越研究中心);School of Cardiovascular & Metabolic Medicine and Sciences, King’s College London(伦敦国王学院心血管与代谢医学与科学学院);Institute of Cardiovascular Science, University College London(伦敦大学学院心血管科学研究所)
该研究在 UK Biobank 大规模生存数据上基准测试八种线性至深度学习生存模型,发现 COX-PH 稳健,而复杂模型仅在特定样本与预测矩阵条件下占优。
An Efficient Training Algorithm for Kernel Survival Support Vector Machines
The Institute of Cancer Research(英国癌症研究所);Technische Universität München(慕尼黑工业大学);Johns Hopkins University(约翰斯·霍普金斯大学);IBM Almaden Research Center(IBM阿尔马登研究中心)
Enhancing Visual Interpretability and Explainability in Functional Survival Trees and Forests
Department of Mathematics and Physics, University of Campania Luigi Vanvitelli(数学与物理系,坎帕尼亚卢igi vanvitelli大学);Department of Economics, Statistics and Business, Universitas Mercatorum(经济学、统计学与商业系,默卡托鲁姆大学)
Deep Survival Analysis for Competing Risk Modeling with Functional Covariates and Missing Data Imputation
Department of Quantitative Health Science, Cleveland Clinic, Cleveland OH, USA(定量健康科学系,克利夫兰诊所);Department of Pulmonary and Critical Care Medicine, Cleveland Clinic, Cleveland OH, USA(呼吸与重症医学科,克利夫兰诊所);Department of Statistics, Purdue University, Lafayette IN, USA(统计系,普渡大学)
Methodology for Comparing Machine Learning Algorithms for Survival Analysis
Instituto Mauá de Tecnologia(毛阿理工学院);Faculdade de Saúde Pública da Universidade de São Paulo(圣保罗大学公共卫生学院);Fundação Oncocentro de São Paulo(圣保罗肿瘤中心基金会);A.C. Camargo Cancer Center(A.C.卡马戈癌症中心);Escola Politécnica da Universidade de São Paulo(圣保罗大学理工学院)
Development of Interactive Nomograms for Predicting Short-Term Survival in ICU Patients with Aplastic Anemia
University of Southern California(南加州大学);Iran University of Medical Sciences(伊朗医科大学);California State University, Long Beach(加州州立大学长滩分校);California State Polytechnic University, Pomona(加州州立理工大学波莫纳分校)
The Australian National University(澳大利亚国立大学);Nanjing University of Science and Technology(南京理工大学);KTH Royal Institute of Technology(瑞典皇家理工学院);The University of Sydney(悉尼大学)
ICODEN: Ordinary Differential Equation Neural Networks for Interval-Censored Data
Department of Biostatistics and Health Data Science, University of Pittsburgh, PA, USA(匹兹堡大学生物统计与健康数据科学系);Center for Applied Statistics and School of Statistics, Renmin University of China, Beijing, China(中国人民大学应用统计中心与统计学院);Department of Applied Statistics, Konkuk University, Seoul, Republic of Korea(韩国康国民大学应用统计系)