Neural Topic Models with Survival Supervision: Jointly Predicting Time-to-Event Outcomes and Learning How Clinical Features Relate
Heinz College of Information Systems and Public Policy, Carnegie Mellon University（卡内基梅隆大学信息系统与公共政策学院）；McKinsey & Company（麦肯锡公司）；Cornerstone Research（基石研究公司）；Microsoft Research（微软研究院）；National Library of Medicine, National Institutes of Health（美国国立卫生研究院国家医学图书馆）
George H. Chen, Linhong Li, Ren Zuo, Amanda Coston, Jeremy C. Weiss
备注：
Accepted at the Artificial Intelligence in Medicine journal; preliminary conference version (see earlier arXiv draft) appeared in the International Conference on Artificial Intelligence in Medicine (AIME 2020)
We present a neural network framework for learning a survival model to predict a time-to-event outcome while simultaneously learning a topic model that reveals feature relationships. In particular, we model each subject as a distribution over "topics", where a topic could, for instance, correspond to an age group, a disorder, or a disease. The presence of a topic in a subject means that specific clinical features are more likely to appear for the subject. Topics encode information about related features and are learned in a supervised manner to predict a time-to-event outcome. Our framework supports combining many different topic and survival models; training the resulting joint survival-topic model readily scales to large datasets using standard neural net optimizers with minibatch gradient descent. For example, a special case is to combine LDA with a Cox model, in which case a subject's distribution over topics serves as the input feature vector to the Cox model. We explain how to address practical implementation issues that arise when applying these neural survival-supervised topic models to clinical data, including how to visualize results to assist clinical interpretation. We study the effectiveness of our proposed framework on seven clinical datasets on predicting time until death as well as hospital ICU length of stay, where we find that neural survival-supervised topic models achieve competitive accuracy with existing approaches while yielding interpretable clinical topics that explain feature relationships. Our code is available at: https://github.com/georgehc/survival-topics
Many clinical studies require the follow-up of patients over time. This is challenging: apart from frequently observed drop-out, there are often also organizational and financial challenges, which can lead to reduced data collection and, in turn, can complicate subsequent analyses. In contrast, there is often plenty of baseline data available of patients with similar characteristics and background information, e.g., from patients that fall outside the study time window. In this article, we investigate whether we can benefit from the inclusion of such unlabeled data instances to predict accurat…
We introduce a new approach for topic modeling that is supervised by survival analysis. Specifically, we build on recent work on unsupervised topic modeling with so-called anchor words by providing supervision through an elastic-net regularized Cox proportional hazards model. In short, an anchor word being present in a document provides strong indication that the document is partially about a specific topic. For example, by seeing "gallstones" in a document, we are fairly certain that the document is partially about medicine. Our proposed method alternates between learning a topic model and le…
University of California, Los Angeles（加州大学洛杉矶分校）；University of Oxford（牛津大学）；Alan Turing Institute（艾伦·图灵研究所）
Survival analysis in the presence of multiple possible adverse events, i.e., competing risks, is a pervasive problem in many industries (healthcare, finance, etc.). Since only one event is typically observed, the incidence of an event of interest is often obscured by other related competing events. This nonidentifiability, or inability to estimate true cause-specific survival curves from empirical data, further complicates competing risk survival analysis. We introduce Siamese Survival Prognosis Network (SSPN), a novel deep learning architecture for estimating personalized risk scores in the p…
NAVER R&D Center（NAVER研发中心）；NAVER CLOVA；NAVER AI LAB（NAVER人工智能实验室）
Survival analysis appears in various fields such as medicine, economics, engineering, and business. Recent studies showed that the Ordinary Differential Equation (ODE) modeling framework unifies many existing survival models while the framework is flexible and widely applicable. However, naively applying the ODE framework to survival analysis problems may model fiercely changing density function which may worsen the model's performance. Though we can apply L1 or L2 regularizers to the ODE model, their effect on the ODE modeling framework is barely known. In this paper, we propose hazard gradie…
Virginia Tech（弗吉尼亚理工大学）；University of Michigan（密歇根大学）
Accurately predicting the time of occurrence of an event of interest is a critical problem in longitudinal data analysis. One of the main challenges in this context is the presence of instances whose event outcomes become unobservable after a certain time point or when some instances do not experience any event during the monitoring period. Such a phenomenon is called censoring which can be effectively handled using survival analysis techniques. Traditionally, statistical approaches have been widely developed in the literature to overcome this censoring issue. In addition, many machine learnin…
Survival analysis has been developed and applied in the number of areas including manufacturing, finance, economics and healthcare. In healthcare domain, usually clinical data are high-dimensional, sparse and complex and sometimes there exists few amount of time-to-event (labeled) instances. Therefore building an accurate survival model from electronic health records is challenging. With this motivation, we address this issue and provide a new survival analysis framework using deep learning and active learning with a novel sampling strategy. First, our approach provides better representation w…
Two challenging problems in the clinical study of cancer are the characterization of cancer subtypes and the classification of individual patients according to those subtypes. Statistical approaches addressing these problems are hampered by population heterogeneity and challenges inherent in data integration across high-dimensional, diverse covariates. We have developed a survival-supervised latent Dirichlet allocation (survLDA) modeling framework to address these concerns. LDA models have proven extremely effective at identifying themes common across large collections of text, but application…
Background: Many authors have described MELD as a predictor of short-term mortality in the liver transplantation waiting list. However MELD score accuracy to predict long term mortality has not been statistically evaluated. Objective: The aim of this study is to analyze the MELD score as well as other variables as a predictor of long-term mortality using a new model: the Survival Tree analysis. Study Design and Setting: The variables obtained at the time of liver transplantation list enrollment and considered in this study are: sex, age, blood type, body mass index, etiology of liver disease, …
University of Kentucky（肯塔基大学）；Qingdao University（青岛大学）
Interpreting critical variables involved in complex biological processes related to survival time can help understand prediction from survival models, evaluate treatment efficacy, and develop new therapies for patients. Currently, the predictive results of deep learning (DL)-based models are better than or as good as standard survival methods, they are often disregarded because of their lack of transparency and little interpretability, which is crucial to their adoption in clinical applications. In this paper, we introduce a novel, easily deployable approach, called EXplainable CEnsored Learni…
School of Electrical and Computer Engineering, Purdue University（普渡大学电气与计算机工程学院）；Department of Medicine, Indiana University School of Medicine（印第安纳大学医学院医学系）；School of Electrical and Computer Engineering, Indiana University - Purdue University Indianapolis（印第安纳大学-普渡大学印第安纳波利斯分校电气与计算机工程学院）
One of the central goals in precision health is the understanding and interpretation of high-dimensional biological data to identify genes and markers associated with disease initiation, development, and outcomes. Though significant effort has been committed to harness gene expression data for multiple analyses while accounting for time-to-event modeling by including survival times, many traditional analyses have focused separately on non-negative matrix factorization (NMF) of the gene expression data matrix and survival regression with Cox proportional hazards model. In this work, Cox proport…
Background and Objective: The survival-agreement plot was proposed and improved to assess the reliability of a quantitative measure. We propose the use of survival analysis as an alternative non-parametric approach for comparison of ordinal qualitative data. Study Design and Setting: Two case studies were presented. The first one is related to a randomized, double blind, placebo-controlled clinical trial to investigate the safety and efficacy of silymarin/metionin for chronic hepatitis C. The second one is a prospective study to identify gustatory alterations due to chorda tympani nerve involv…
Proportional hazards (PH), proportional odds (PO) and accelerated failure time (AFT) models have been widely used to deal with survival data in different fields of knowledge. Despite their popularity, such models are not suitable to handle survival data with crossing survival curves. Yang and Prentice (2005) proposed a semiparametric two-sample approach, denoted here as the YP model, allowing the analysis of crossing survival curves and including the PH and PO configurations as particular cases. In a general regression setting, the present work proposes a fully likelihood-based approach to fit…
Extended cure survival models enable to separate covariates that affect the probability of an event (or `long-term' survival) from those only affecting the event timing (or `short-term' survival). We propose to generalize the bounded cumulative hazard model to handle additive terms for time-varying (exogenous) covariates jointly impacting long- and short-term survival. The selection of the penalty parameters is a challenge in that framework. A fast algorithm based on Laplace approximations in Bayesian P-spline models is proposed. The methodology is motivated by fertility studies where women's …
Neural interval-censored survival regression with feature selection
Instituto de Ciencias Matemáticas, CSIC（数学科学研究所（西班牙高等科学研究理事会））；Universidad Autónoma de Madrid（马德里自治大学）；Harvard University（哈佛大学）；Université Paris 1 Panthéon-Sorbonne（巴黎第一大学先贤祠-索邦大学）；University of North Carolina, Chapel Hill（北卡罗来纳大学教堂山分校）
The University of Iowa（爱荷华大学）；Worcester Polytechnic Institute（伍斯特理工学院）
Urban dispersal events are processes where an unusually large number of people leave the same area in a short period. Early prediction of dispersal events is important in mitigating congestion and safety risks and making better dispatching decisions for taxi and ride-sharing fleets. Existing work mostly focuses on predicting taxi demand in the near future by learning patterns from historical data. However, they fail in case of abnormality because dispersal events with abnormally high demand are non-repetitive and violate common assumptions such as smoothness in demand change over time. Instead…
Several gene-based association tests for time-to-event traits have been proposed recently, to detect whether a gene region (containing multiple variants), as a set, is associated with the survival outcome. However, for bivariate survival outcomes, to the best of our knowledge, there is no statistical method that can be directly applied for gene-based association analysis. Motivated by a genetic study to discover gene regions associated with the progression of a bilateral eye disease, Age-related Macular Degeneration (AMD), we implement a novel functional regression method under the copula fram…
Illness-death models are a class of stochastic models inside the multi-state framework. In those models, individuals are allowed to move over time between different states related to illness and death. They are of special interest when working with non-terminal diseases, as they not only consider the competing risk of death but also allow to study progression from illness to death. The intensity of each transition can be modelled including both fixed and random effects of covariates. In particular, spatially structured random effects or their multivariate versions can be used to assess spatial…
The proportional hazards (PH), proportional odds (PO) and accelerated failure time (AFT) models have been widely used in different applications of survival analysis. Despite their popularity, these models are not suitable to handle lifetime data with crossing survival curves. In 2005, Yang and Prentice proposed a semiparametric two-sample strategy (YP model), including the PH and PO frameworks as particular cases, to deal with this type of data. Assuming a general regression setting, the present paper proposes an unified approach to fit the YP model by employing Bernstein polynomials to manage…
What Risk Factors to Cause Long COVID and Its Impact on Patient Survival Outcomes when Combined with the Effect from Organ Transplantation in the Acute COVID
Patient-level health economic data collected alongside clinical trials are an important component of the process of technology appraisal, with a view to informing resource allocation decisions. For end of life treatments, such as cancer treatments, modelling of cost-effectiveness/utility data may involve some form of partitioned survival analysis, where measures of health-related quality of life and survival time for both pre- and post-progression periods are combined to generate some aggregate measure of clinical benefits (e.g. quality-adjusted survival). In addition, resource use data are of…
The Roman Empire shaped Western civilization, and many Roman principles are embodied in modern institutions. Although its political institutions proved both resilient and adaptable, allowing it to incorporate diverse populations, the Empire suffered from many internal conflicts. Indeed, most emperors died violently, from assassination, suicide, or in battle. These internal conflicts produced patterns in the length of time that can be identified by statistical analysis. In this paper, we study the underlying patterns associated with the reign of the Roman emperors by using statistical tools of …
Predictive modeling from high-dimensional genomic data is often preceded by a dimension reduction step, such as principal components analysis (PCA). However, the application of PCA is not straightforward for multi-source data, wherein multiple sources of 'omics data measure different but related biological components. In this article we utilize recent advances in the dimension reduction of multi-source data for predictive modeling. In particular, we apply exploratory results from Joint and Individual Variation Explained (JIVE), an extension of PCA for multi-source data, for prediction of diffe…
Southern Methodist University（南卫理公会大学）；University of Texas at Dallas（德克萨斯大学达拉斯分校）；University of Texas Southwestern Medical Center（德克萨斯大学西南医学中心）
Tumor shape is a key factor that affects tumor growth and metastasis. This paper proposes a topological feature computed by persistent homology to characterize tumor progression from digital pathology and radiology images and examines its effect on the time-to-event data. The proposed topological features are invariant to scale-preserving transformation and can summarize various tumor shape patterns. The topological features are represented in functional space and used as functional predictors in a functional Cox proportional hazards model. The proposed model enables interpretable inference ab…
In reliability theory and survival analysis, the residual entropy is known as a measure suitable to describe the dynamic information content in stochastic systems conditional on survival. Aiming to analyze the variability of such information content, in this paper we introduce the variance of the residual lifetimes, "residual varentropy" in short. After a theoretical investigation of some properties of the residual varentropy, we illustrate certain applications related to the proportional hazards model and the first-passage times of an Ornstein-Uhlenbeck jump-diffusion process.
ETH Zürich（苏黎世联邦理工学院）；Human Technopole（人类技术园）；University Hospital Basel（巴塞尔大学医院）；University Children’s Hospital Basel（巴塞尔大学儿童医院）；University of Zürich（苏黎世大学）；Politecnico di Milano（米兰理工大学）；St. Gallen Cantonal Hospital（圣加仑州立医院）
In this work, we study the problem of clustering survival data $-$ a challenging and so far under-explored task. We introduce a novel semi-supervised probabilistic approach to cluster survival data by leveraging recent advances in stochastic gradient variational inference. In contrast to previous work, our proposed method employs a deep generative model to uncover the underlying distribution of both the explanatory variables and censored survival times. We compare our model to the related work on clustering and mixture models for survival data in comprehensive experiments on a wide range of sy…
Deep learning excels in the analysis of unstructured data and recent advancements allow to extend these techniques to survival analysis. In the context of clinical radiology, this enables, e.g., to relate unstructured volumetric images to a risk score or a prognosis of life expectancy and support clinical decision making. Medical applications are, however, associated with high criticality and consequently, neither medical personnel nor patients do usually accept black box models as reason or basis for decisions. Apart from averseness to new technologies, this is due to missing interpretability…
AdaMSS: Adaptive Multi-Modality Segmentation-to-Survival Learning for Survival Outcome Prediction from PET/CT Images
The University of Sydney（悉尼大学）；Shanghai Jiao Tong University（上海交通大学）；Fudan University Shanghai Cancer Center（复旦大学附属肿瘤医院）；Fudan University（复旦大学）；Royal Prince Alfred Hospital（皇家阿尔弗雷德亲王医院）
Data-driven approaches which aim to identify and predict player engagement are becoming increasingly popular in games industry contexts. This is due to the growing practice of tracking and storing large volumes of in-game telemetries coupled with a desire to tailor the gaming experience to the end-user's needs. These approaches are particularly useful not just for companies adopting Game-as-a-Service (GaaS) models (e.g. for re-engagement strategies) but also for those working under persistent content-delivery regimes (e.g. for better audience targeting). A major challenge for the latter is to …
Predicting Patient Survival with Airway Biomarkers using nn-Unet/Radiomics
Normandie Univ., INSA Rouen, LITIS, 76801, Saint Etienne du Rouvray, France（诺曼底大学）；Normandie Univ., UNICAEN, ENSICAEN, CNRS, GREYC, 14000 Caen, France（诺曼底大学）；Nuclear Medicine Department, Henri Becquerel Cancer Center, Rouen, France（贝克勒尔癌症中心核医学部）；Radiotherapy Department, Henri Becquerel Cancer Center, Rouen, France（贝克勒尔癌症中心放射治疗部）；Siemens Healthineers（西门子医疗）
SELECTOR: Heterogeneous graph network with convolutional masked autoencoder for multimodal robust prediction of cancer survival
College of Computer Science and Electronic Engineering, Hunan University（湖南大学计算机与通信工程学院）；The Second Xiangya Hospital, Central South University（中南大学湘雅二医院）
Idiopathic Pulmonary Fibrosis (IPF) is an inexorably progressive fibrotic lung disease with a variable and unpredictable rate of progression. CT scans of the lungs inform clinical assessment of IPF patients and contain pertinent information related to disease progression. In this work, we propose a multi-modal method that uses neural networks and memory banks to predict the survival of IPF patients using clinical and imaging data. The majority of clinical IPF patient records have missing data (e.g. missing lung function tests). To this end, we propose a probabilistic model that captures the de…
University of Iowa（爱荷华大学）；Stanford University（斯坦福大学）
Non-small-cell lung cancer (NSCLC) represents approximately 80-85% of lung cancer diagnoses and is the leading cause of cancer-related death worldwide. Recent studies indicate that image-based radiomics features from positron emission tomography-computed tomography (PET/CT) images have predictive power on NSCLC outcomes. To this end, easily calculated functional features such as the maximum and the mean of standard uptake value (SUV) and total lesion glycolysis (TLG) are most commonly used for NSCLC prognostication, but their prognostic value remains controversial. Meanwhile, convolutional neu…
Supervised learning algorithms are heavily reliant on annotated datasets to train machine learning models. However, the curation of the annotated datasets is laborious and time consuming due to the manual effort involved and has become a huge bottleneck in supervised learning. In this work, we apply the theory of noisy learning to generate weak supervision signals instead of manual annotation. We curate a noisy labeled dataset using a labeling heuristic to identify epidemic related tweets. We evaluated the performance using a large epidemic corpus and our results demonstrate that models traine…
Imperial College London（帝国理工学院）；Samsung AI Centre, Cambridge（三星人工智能中心（剑桥））；Facebook London（Facebook（伦敦））
The intuitive interaction between the audio and visual modalities is valuable for cross-modal self-supervised learning. This concept has been demonstrated for generic audiovisual tasks like video action recognition and acoustic scene classification. However, self-supervision remains under-explored for audiovisual speech. We propose a method to learn self-supervised speech representations from the raw audio waveform. We train a raw audio encoder by combining audio-only self-supervision (by predicting informative audio attributes) with visual self-supervision (by generating talking faces from au…
