Machine Learning for Survival Analysis

Wang, Ping; Li, Yan; Reddy, Chandan K.

doi:10.1145/3214306

Cited by 538 publications

(354 citation statements)

References 114 publications

Supporting

Mentioning

346

Contrasting

Unclassified

Order By: Relevance

“…When the total number of samples is large, we can roughly assume h(t) to be close enough at the time of CV. The comparison of h i (t) thus reduces to the comparison of risk score ⌘ i = | X i , i.e., the logarithm of "hazard ratio" exp ( | X i ), which is independent of time t [78].…”

Section: Cox Regression and Evaluationmentioning

confidence: 99%

Improving personalized prediction of cancer prognoses with clonal evolution models

Tao

Rajaraman

Cui

et al. 2019

Preprint

View full text Add to dashboard Cite

Cancer occurs via an accumulation of somatic genomic alterations in a process of clonal evolution. There has been intensive study of potential causal mutations driving cancer development and progression. However, much recent evidence suggests that tumor evolution is normally driven by a variety of mechanisms of somatic hypermutability, known as mutator phenotypes, which act in different combinations or degrees in different cancers. Here we explore the question of how and to which degree different mutator phenotypes act in a cancer predict its future progression. We develop a computational paradigm using evolutionary tree inference (tumor phylogeny) algorithms to derive features quantifying single-tumor mutational preferences, followed by a machine learning framework to identify key features predictive of progression. We build phylogenies tracing the evolution of subclones of cells in tumor tissues using a variety of somatic genomic alterations, including single nucleotide variations, copy number alterations, and structural variations. We demonstrate that mutation preference features derived from the phylogenies are predictive of clinical outcomes of cancer progression -overall survival and disease-free survival -based on the analyses on breast invasive carcinoma, lung adenocarcinoma, and lung squamous cell carcinoma. We further show that mutational phenotypes have predictive power even after accounting for traditional clinical and driver-centric predictors of progression. These results confirm the power of mutational phenotypes as an independent class of predictive biomarkers and suggest a strategy for enhancing the predictive power of conventional clinical or driver-centric genomic features.

show abstract

Section: Cox Regression and Evaluationmentioning

confidence: 99%

Improving personalized prediction of cancer prognoses with clonal evolution models

Tao

Rajaraman

Cui

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Survival analysis has traditionally been used in the health-care domain to determine the time to 'death' in patients, but the usage of this range of techniques has recently expanded to other application areas [37]. Examples include prediction of early student dropouts [1], post-click engagement on native ads [2], query specific microblog ranking for improved retrieval [17], recommender systems in e-commerce [36], search engine evaluation via the use of "absence time" [7], and predicting time for crowd-sourced tasks [26].…”

Section: Related Workmentioning

confidence: 99%

Modeling Time to Open of Emails with a Latent State for User Engagement Level

Sinha¹,

Vinay²,

Singh³

2018

Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining

View full text Add to dashboard Cite

Email messages have been an important mode of communication, not only for work, but also for social interactions and marketing. When messages have time sensitive information, it becomes relevant for the sender to know what is the expected time within which the email will be read by the recipient. In this paper we use a survival analysis framework to predict the time to open an email once it has been received. We use the Cox Proportional Hazards (CoxPH) model that offers a way to combine various features that might affect the event of opening an email. As an extension, we also apply a mixture model (MM) approach to CoxPH that distinguishes between recipients, based on a latent state of how prone to opening the messages each individual is. We compare our approach with standard classification and regression models. While the classification model provides predictions on the likelihood of an email being opened, the regression model provides prediction of the real-valued time to open. The use of survival analysis based methods allows us to jointly model both the open event as well as the time-to-open. We experimented on a large real-world dataset of marketing emails sent in a 3-month time duration. The mixture model achieves the best accuracy on our data where a high proportion of email messages go unopened.

show abstract

“…While vastly popular among practitioners, these models have been criticized for a number of reasons, in particular for the assumptions they make, that consequently render them unfit for many modern applications [45]. For instance, most survival models, including CoxPH and the proportional odds model [31], work under the premise of fixed covariate effects, overlooking individual uncertainty.…”

Section: Introductionmentioning

confidence: 99%

Variational learning of individual survival distributions

Xiu

Tao

Henao

2020

Proceedings of the ACM Conference on Health, Inference, and Learning

View full text Add to dashboard Cite

The abundance of modern health data provides many opportunities for the use of machine learning techniques to build better statistical models to improve clinical decision making. Predicting time-toevent distributions, also known as survival analysis, plays a key role in many clinical applications. We introduce a variational timeto-event prediction model, named Variational Survival Inference (VSI), which builds upon recent advances in distribution learning techniques and deep neural networks. VSI addresses the challenges of non-parametric distribution estimation by (i) relaxing the restrictive modeling assumptions made in classical models, and (ii) efficiently handling the censored observations, i.e., events that occur outside the observation window, all within the variational framework. To validate the effectiveness of our approach, an extensive set of experiments on both synthetic and real-world datasets is carried out, showing improved performance relative to competing solutions.

show abstract

Machine Learning for Survival Analysis

Cited by 538 publications

References 114 publications

Improving personalized prediction of cancer prognoses with clonal evolution models

Improving personalized prediction of cancer prognoses with clonal evolution models

Modeling Time to Open of Emails with a Latent State for User Engagement Level

Variational learning of individual survival distributions

Contact Info

Product

Resources

About