Censored imputation of time to event outcome through survival proximity score method

TY - JOUR

T1 - Censored imputation of time to event outcome through survival proximity score method

AU - Vishwakarma, Gajendra K.

AU - Bhattacharjee, Atanu

AU - Rajbongshi, Bhrigu K.

AU - Tripathy, Abhipsa

N1 - Copyright: © 2024 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

PY - 2024/12/1

Y1 - 2024/12/1

N2 - The study recognized the significance of accurately addressing censoring and censored data in survival analysis as well as the possibility of biases if these concepts are not approached with caution by researchers. If censored participants are excluded from the study, the true treatment effect comparison will be inaccurate, resulting in biased study findings. In this context, there is a need for some mechanism to impute and update the status information of those patients who were lost to follow-up before the study’s end and might be due to the study-related causes. In order to resolve this issue, we developed a weighted survival proximity score method by using a weighted propensity scoring system to impute the censored values with the help of the nearest neighbourhood technique and update the survival data. After imputing the censored status in the dataset, conventional survival methods are applied to evaluate the differences in hazard rates with and without applying a survival proximity score method. We found that the proposed method reduces bias and mean square error in various transitions. The goal of this work is to resolve the existing issue of censoring and explain how the survival proximity score method can be adapted to update the censored information of lost-followup individuals for therapeutic arms comparison.

AB - The study recognized the significance of accurately addressing censoring and censored data in survival analysis as well as the possibility of biases if these concepts are not approached with caution by researchers. If censored participants are excluded from the study, the true treatment effect comparison will be inaccurate, resulting in biased study findings. In this context, there is a need for some mechanism to impute and update the status information of those patients who were lost to follow-up before the study’s end and might be due to the study-related causes. In order to resolve this issue, we developed a weighted survival proximity score method by using a weighted propensity scoring system to impute the censored values with the help of the nearest neighbourhood technique and update the survival data. After imputing the censored status in the dataset, conventional survival methods are applied to evaluate the differences in hazard rates with and without applying a survival proximity score method. We found that the proposed method reduces bias and mean square error in various transitions. The goal of this work is to resolve the existing issue of censoring and explain how the survival proximity score method can be adapted to update the censored information of lost-followup individuals for therapeutic arms comparison.

KW - Multistate model

KW - Cox proportional hazard model

KW - Censoring

KW - Survival proximity score matching

KW - Weights

KW - Data simulation

UR - https://www.scopus.com/pages/publications/85196780764

U2 - 10.1016/j.cam.2024.116103

DO - 10.1016/j.cam.2024.116103

M3 - Article

SN - 0377-0427

VL - 451

JO - Journal of Computational and Applied Mathematics

JF - Journal of Computational and Applied Mathematics

M1 - 116103

ER -