Enabling counterfactual survival analysis with balanced representations

Chapfuwa, Paidamoyo; Assaad, Serge; Zeng, Shuxi; Pencina, Michael J.; Carin, Lawrence; Henao, Ricardo

doi:10.1145/3450439.3451875

Cited by 17 publications

(71 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The recently proposed continuous-time SA based counterfactual survival analysis (CSA) [33] and discrete-time SA based SurvITE [34] are the only techniques for heterogeneous and personalized CI in SA datasets. Both these models optimize the average (heterogeneous) treatment effect via counterfactual prediction of survival outcomes in observational studies, while leveraging on the balanced representation learning [6].…”

Section: Related Work and Noveltymentioning

confidence: 99%

“…• ACTG-Semi synthetic dataset: The AIDS Clinical Trials Group (ACTG) is an RCT study [42]. Generation of the semi-synthetic dataset based on covariates from ACTG is as described in [33]. Generation of the semi-synthetic dataset based on covariates from ACTG is as described in [33].…”

Section: A Datasetsmentioning

confidence: 99%

“…Generation of the semi-synthetic dataset based on covariates from ACTG is as described in [33]. Generation of the semi-synthetic dataset based on covariates from ACTG is as described in [33]. The time-to-censoring for the non-informative censoring scenario is generated as log(Y C,i ) ∼ N (µ c , σ 2 c ).…”

Section: A Datasetsmentioning

confidence: 99%

See 2 more Smart Citations

Learn to Live Longer: Counterfactual Inference using Balanced Representations for Parametric Deep Survival Analysis

Prasad¹

2023

Preprint

View full text Add to dashboard Cite

show abstract

Section: Related Work and Noveltymentioning

confidence: 99%

Section: A Datasetsmentioning

confidence: 99%

See 1 more Smart Citation

Learn to Live Longer: Counterfactual Inference using Balanced Representations for Parametric Deep Survival Analysis

Prasad¹

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Potential outcomes can then be estimated by changing the respective treatment covariate or model. These naive approaches are occasionally discussed in performance comparisons, e.g., in (Chapfuwa et al, 2020;Curth et al, 2021). An alternative approach is to match similar patients between treated and non-treated populations using, e.g., propensity scores (Rosenbaum and Rubin, 1983).…”

Section: Introductionmentioning

confidence: 99%

BITES: Balanced Individual Treatment Effect for Survival data

Schrod¹,

Schäfer²,

Solbrig³

et al. 2022

Preprint

View full text Add to dashboard Cite

Estimating the effects of interventions on patient outcome is one of the key aspects of personalized medicine. Their inference is often challenged by the fact that the training data comprises only the outcome for the administered treatment, and not for alternative treatments (the so-called counterfactual outcomes). Several methods were suggested for this scenario based on observational data, i.e. data where the intervention was not applied randomly, for both continuous and binary outcome variables. However, patient outcome is often recorded in terms of time-to-event data, comprising right-censored event times if an event does not occur within the observation period. Albeit their enormous importance, time-to-event data is rarely used for treatment optimization.We suggest an approach named BITES (Balanced Individual Treatment Effect for Survival data), which combines a treatment-specific semiparametric Cox loss with a treatment-balanced deep neural network; i.e. we regularize differences between treated and non-treated patients using Integral Probability Metrics (IPM). We show in simulation studies that this approach outperforms the state of the art. Further, we demonstrate in an application to a cohort of breast cancer patients that hormone treatment can be optimized based on six routine parameters. We successfully validated this finding in an independent cohort.

show abstract

“…Nonparametrically investigating treatment effect heterogeneity, however, has been studied in much less detail in the survival context. While a number of tree-based methods have been proposed recently [23,24,25,26], NN-based methods lack extensions to the time-to-event setting despite their successful adoption for estimating the effects of treatments on other outcomes -the only exception being [27], who directly model event times under different treatments with generative models.…”

Section: Introductionmentioning

confidence: 99%

SurvITE: Learning Heterogeneous Treatment Effects from Time-to-Event Data

Curth¹,

Lee²,

Schaar³

2021

Preprint

View full text Add to dashboard Cite

We study the problem of inferring heterogeneous treatment effects from time-toevent data. While both the related problems of (i) estimating treatment effects for binary or continuous outcomes and (ii) predicting survival outcomes have been well studied in the recent machine learning literature, their combination -albeit of high practical relevance -has received considerably less attention. With the ultimate goal of reliably estimating the effects of treatments on instantaneous risk and survival probabilities, we focus on the problem of learning (discrete-time) treatment-specific conditional hazard functions. We find that unique challenges arise in this context due to a variety of covariate shift issues that go beyond a mere combination of wellstudied confounding and censoring biases. We theoretically analyse their effects by adapting recent generalization bounds from domain adaptation and treatment effect estimation to our setting and discuss implications for model design. We use the resulting insights to propose a novel deep learning method for treatment-specific hazard estimation based on balancing representations. We investigate performance across a range of experimental settings and empirically confirm that our method outperforms baselines by addressing covariate shifts from various sources.

show abstract

Enabling counterfactual survival analysis with balanced representations

Cited by 17 publications

References 19 publications

Learn to Live Longer: Counterfactual Inference using Balanced Representations for Parametric Deep Survival Analysis

Learn to Live Longer: Counterfactual Inference using Balanced Representations for Parametric Deep Survival Analysis

BITES: Balanced Individual Treatment Effect for Survival data

SurvITE: Learning Heterogeneous Treatment Effects from Time-to-Event Data

Contact Info

Product

Resources

About