Neural network analysis to predict mortality in end-stage renal disease

Application to united states renal data system

Adam N. Jacob, Sadik Khuder, Nathan Malhotra, Thomas Sodeman, Jeffrey P Gold, Deepak Malhotra, Joseph I. Shapiro

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We examined whether we could develop models based on data provided to the United States Renal Data System (USRDS) to accurately predict survival. Records were obtained from patients beginning dialysis in 1990 through 2007. We developed linear and neural network models and optimized the fit of these models to the actual time to death. Next, we examined whether we could accurately predict survival in a dataset containing censored and uncensored patients. The results with these models were contrasted with those obtained with a Cox proportional hazards model fit to the entire dataset. The average C statistic over a 6-month to 10-year time range achieved with these models was approximately 0.7891 (linear model), 0.7804 (transformed dataset linear model), 0.7769 (neural network model), 0.7774 (transformed dataset neural network model), 0.8019 (Cox model), and 0.7970 (transformed dataset Cox model). When we used the Cox proportional hazards model, superior C statistic results were found at time points between 2 and 10 years but at earlier time points, the Cox model was slightly inferior. These results suggest that data provided to the USRDS can allow for predictive models which have a high degree of accuracy years following the initiation of dialysis.

Original languageEnglish (US)
JournalNephron - Clinical Practice
Volume116
Issue number2
DOIs
StatePublished - Sep 1 2010

Fingerprint

Proportional Hazards Models
Information Systems
Chronic Kidney Failure
Neural Networks (Computer)
Kidney
Mortality
Dialysis
Linear Models
Survival
Datasets

Keywords

  • End-stage renal disease
  • Medicare
  • Mortality
  • Neural network
  • Renal failure
  • United States Renal Data System

ASJC Scopus subject areas

  • Nephrology

Cite this

Neural network analysis to predict mortality in end-stage renal disease : Application to united states renal data system. / Jacob, Adam N.; Khuder, Sadik; Malhotra, Nathan; Sodeman, Thomas; Gold, Jeffrey P; Malhotra, Deepak; Shapiro, Joseph I.

In: Nephron - Clinical Practice, Vol. 116, No. 2, 01.09.2010.

Research output: Contribution to journalArticle

Jacob, Adam N. ; Khuder, Sadik ; Malhotra, Nathan ; Sodeman, Thomas ; Gold, Jeffrey P ; Malhotra, Deepak ; Shapiro, Joseph I. / Neural network analysis to predict mortality in end-stage renal disease : Application to united states renal data system. In: Nephron - Clinical Practice. 2010 ; Vol. 116, No. 2.
@article{1c52699a96a64d3ca22676b0fa6331d0,
title = "Neural network analysis to predict mortality in end-stage renal disease: Application to united states renal data system",
abstract = "We examined whether we could develop models based on data provided to the United States Renal Data System (USRDS) to accurately predict survival. Records were obtained from patients beginning dialysis in 1990 through 2007. We developed linear and neural network models and optimized the fit of these models to the actual time to death. Next, we examined whether we could accurately predict survival in a dataset containing censored and uncensored patients. The results with these models were contrasted with those obtained with a Cox proportional hazards model fit to the entire dataset. The average C statistic over a 6-month to 10-year time range achieved with these models was approximately 0.7891 (linear model), 0.7804 (transformed dataset linear model), 0.7769 (neural network model), 0.7774 (transformed dataset neural network model), 0.8019 (Cox model), and 0.7970 (transformed dataset Cox model). When we used the Cox proportional hazards model, superior C statistic results were found at time points between 2 and 10 years but at earlier time points, the Cox model was slightly inferior. These results suggest that data provided to the USRDS can allow for predictive models which have a high degree of accuracy years following the initiation of dialysis.",
keywords = "End-stage renal disease, Medicare, Mortality, Neural network, Renal failure, United States Renal Data System",
author = "Jacob, {Adam N.} and Sadik Khuder and Nathan Malhotra and Thomas Sodeman and Gold, {Jeffrey P} and Deepak Malhotra and Shapiro, {Joseph I.}",
year = "2010",
month = "9",
day = "1",
doi = "10.1159/000315884",
language = "English (US)",
volume = "116",
journal = "Nephron - Experimental Nephrology",
issn = "1660-8151",
publisher = "S. Karger AG",
number = "2",

}

TY - JOUR

T1 - Neural network analysis to predict mortality in end-stage renal disease

T2 - Application to united states renal data system

AU - Jacob, Adam N.

AU - Khuder, Sadik

AU - Malhotra, Nathan

AU - Sodeman, Thomas

AU - Gold, Jeffrey P

AU - Malhotra, Deepak

AU - Shapiro, Joseph I.

PY - 2010/9/1

Y1 - 2010/9/1

N2 - We examined whether we could develop models based on data provided to the United States Renal Data System (USRDS) to accurately predict survival. Records were obtained from patients beginning dialysis in 1990 through 2007. We developed linear and neural network models and optimized the fit of these models to the actual time to death. Next, we examined whether we could accurately predict survival in a dataset containing censored and uncensored patients. The results with these models were contrasted with those obtained with a Cox proportional hazards model fit to the entire dataset. The average C statistic over a 6-month to 10-year time range achieved with these models was approximately 0.7891 (linear model), 0.7804 (transformed dataset linear model), 0.7769 (neural network model), 0.7774 (transformed dataset neural network model), 0.8019 (Cox model), and 0.7970 (transformed dataset Cox model). When we used the Cox proportional hazards model, superior C statistic results were found at time points between 2 and 10 years but at earlier time points, the Cox model was slightly inferior. These results suggest that data provided to the USRDS can allow for predictive models which have a high degree of accuracy years following the initiation of dialysis.

AB - We examined whether we could develop models based on data provided to the United States Renal Data System (USRDS) to accurately predict survival. Records were obtained from patients beginning dialysis in 1990 through 2007. We developed linear and neural network models and optimized the fit of these models to the actual time to death. Next, we examined whether we could accurately predict survival in a dataset containing censored and uncensored patients. The results with these models were contrasted with those obtained with a Cox proportional hazards model fit to the entire dataset. The average C statistic over a 6-month to 10-year time range achieved with these models was approximately 0.7891 (linear model), 0.7804 (transformed dataset linear model), 0.7769 (neural network model), 0.7774 (transformed dataset neural network model), 0.8019 (Cox model), and 0.7970 (transformed dataset Cox model). When we used the Cox proportional hazards model, superior C statistic results were found at time points between 2 and 10 years but at earlier time points, the Cox model was slightly inferior. These results suggest that data provided to the USRDS can allow for predictive models which have a high degree of accuracy years following the initiation of dialysis.

KW - End-stage renal disease

KW - Medicare

KW - Mortality

KW - Neural network

KW - Renal failure

KW - United States Renal Data System

UR - http://www.scopus.com/inward/record.url?scp=77952826769&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77952826769&partnerID=8YFLogxK

U2 - 10.1159/000315884

DO - 10.1159/000315884

M3 - Article

VL - 116

JO - Nephron - Experimental Nephrology

JF - Nephron - Experimental Nephrology

SN - 1660-8151

IS - 2

ER -