Machine Learning to Identify Dialysis Patients at High Death Risk

Oguz Akbilgic, Yoshitsugu Obi, Praveen K. Potukuchi, Ibrahim Karabayir, Danh V. Nguyen, Melissa Soohoo, Elani Streja, Miklos Z. Molnar, Connie M. Rhee, Kamyar Kalantar-Zadeh, Csaba Kovesdy

Research output: Contribution to journalArticle

Abstract

Introduction: Given the high mortality rate within the first year of dialysis initiation, an accurate estimation of postdialysis mortality could help patients and clinicians in decision making about initiation of dialysis. We aimed to use machine learning (ML) by incorporating complex information from electronic health records to predict patients at risk for postdialysis short-term mortality. Methods: This study was carried out on a contemporary cohort of 27,615 US veterans with incident end-stage renal disease (ESRD). We implemented a random forest method on 49 variables obtained before dialysis transition to predict outcomes of 30-, 90-, 180-, and 365-day all-cause mortality after dialysis initiation. Results: The mean (±SD) age of our cohort was 68.7 ± 11.2 years, 98.1% of patients were men, 29.4% were African American, and 71.4% were diabetic. The final random forest model provided C-statistics (95% confidence intervals) of 0.7185 (0.6994–0.7377), 0.7446 (0.7346–0.7546), 0.7504 (0.7425–0.7583), and 0.7488 (0.7421–0.7554) for predicting risk of death within the 4 different time windows. The models showed good internal validity and replicated well in patients with various demographic and clinical characteristics and provided similar or better performance compared with other ML algorithms. Results may not be generalizable to non-veterans. Use of predictors available in electronic medical records has limited the assessment of number of predictors. Conclusion: We implemented and ML-based method to accurately predict short-term postdialysis mortality in patients with incident ESRD. Our models could aid patients and clinicians in better decision making about the best course of action in patients approaching ESRD.

Original languageEnglish (US)
Pages (from-to)1219-1229
Number of pages11
JournalKidney International Reports
Volume4
Issue number9
DOIs
StatePublished - Sep 1 2019

Fingerprint

Chronic Renal Insufficiency
Chronic Kidney Failure
Dialysis
Mortality
Electronic Health Records
Decision Making
Veterans
Machine Learning
African Americans
Demography
Confidence Intervals

All Science Journal Classification (ASJC) codes

  • Nephrology

Cite this

Machine Learning to Identify Dialysis Patients at High Death Risk. / Akbilgic, Oguz; Obi, Yoshitsugu; Potukuchi, Praveen K.; Karabayir, Ibrahim; Nguyen, Danh V.; Soohoo, Melissa; Streja, Elani; Molnar, Miklos Z.; Rhee, Connie M.; Kalantar-Zadeh, Kamyar; Kovesdy, Csaba.

In: Kidney International Reports, Vol. 4, No. 9, 01.09.2019, p. 1219-1229.

Research output: Contribution to journalArticle

Akbilgic, O, Obi, Y, Potukuchi, PK, Karabayir, I, Nguyen, DV, Soohoo, M, Streja, E, Molnar, MZ, Rhee, CM, Kalantar-Zadeh, K & Kovesdy, C 2019, 'Machine Learning to Identify Dialysis Patients at High Death Risk', Kidney International Reports, vol. 4, no. 9, pp. 1219-1229. https://doi.org/10.1016/j.ekir.2019.06.009
Akbilgic, Oguz ; Obi, Yoshitsugu ; Potukuchi, Praveen K. ; Karabayir, Ibrahim ; Nguyen, Danh V. ; Soohoo, Melissa ; Streja, Elani ; Molnar, Miklos Z. ; Rhee, Connie M. ; Kalantar-Zadeh, Kamyar ; Kovesdy, Csaba. / Machine Learning to Identify Dialysis Patients at High Death Risk. In: Kidney International Reports. 2019 ; Vol. 4, No. 9. pp. 1219-1229.
@article{38cc6c1da0af47818cf67fae36e2dd60,
title = "Machine Learning to Identify Dialysis Patients at High Death Risk",
abstract = "Introduction: Given the high mortality rate within the first year of dialysis initiation, an accurate estimation of postdialysis mortality could help patients and clinicians in decision making about initiation of dialysis. We aimed to use machine learning (ML) by incorporating complex information from electronic health records to predict patients at risk for postdialysis short-term mortality. Methods: This study was carried out on a contemporary cohort of 27,615 US veterans with incident end-stage renal disease (ESRD). We implemented a random forest method on 49 variables obtained before dialysis transition to predict outcomes of 30-, 90-, 180-, and 365-day all-cause mortality after dialysis initiation. Results: The mean (±SD) age of our cohort was 68.7 ± 11.2 years, 98.1{\%} of patients were men, 29.4{\%} were African American, and 71.4{\%} were diabetic. The final random forest model provided C-statistics (95{\%} confidence intervals) of 0.7185 (0.6994–0.7377), 0.7446 (0.7346–0.7546), 0.7504 (0.7425–0.7583), and 0.7488 (0.7421–0.7554) for predicting risk of death within the 4 different time windows. The models showed good internal validity and replicated well in patients with various demographic and clinical characteristics and provided similar or better performance compared with other ML algorithms. Results may not be generalizable to non-veterans. Use of predictors available in electronic medical records has limited the assessment of number of predictors. Conclusion: We implemented and ML-based method to accurately predict short-term postdialysis mortality in patients with incident ESRD. Our models could aid patients and clinicians in better decision making about the best course of action in patients approaching ESRD.",
author = "Oguz Akbilgic and Yoshitsugu Obi and Potukuchi, {Praveen K.} and Ibrahim Karabayir and Nguyen, {Danh V.} and Melissa Soohoo and Elani Streja and Molnar, {Miklos Z.} and Rhee, {Connie M.} and Kamyar Kalantar-Zadeh and Csaba Kovesdy",
year = "2019",
month = "9",
day = "1",
doi = "10.1016/j.ekir.2019.06.009",
language = "English (US)",
volume = "4",
pages = "1219--1229",
journal = "Kidney International Reports",
issn = "2468-0249",
publisher = "Elsevier Inc.",
number = "9",

}

TY - JOUR

T1 - Machine Learning to Identify Dialysis Patients at High Death Risk

AU - Akbilgic, Oguz

AU - Obi, Yoshitsugu

AU - Potukuchi, Praveen K.

AU - Karabayir, Ibrahim

AU - Nguyen, Danh V.

AU - Soohoo, Melissa

AU - Streja, Elani

AU - Molnar, Miklos Z.

AU - Rhee, Connie M.

AU - Kalantar-Zadeh, Kamyar

AU - Kovesdy, Csaba

PY - 2019/9/1

Y1 - 2019/9/1

N2 - Introduction: Given the high mortality rate within the first year of dialysis initiation, an accurate estimation of postdialysis mortality could help patients and clinicians in decision making about initiation of dialysis. We aimed to use machine learning (ML) by incorporating complex information from electronic health records to predict patients at risk for postdialysis short-term mortality. Methods: This study was carried out on a contemporary cohort of 27,615 US veterans with incident end-stage renal disease (ESRD). We implemented a random forest method on 49 variables obtained before dialysis transition to predict outcomes of 30-, 90-, 180-, and 365-day all-cause mortality after dialysis initiation. Results: The mean (±SD) age of our cohort was 68.7 ± 11.2 years, 98.1% of patients were men, 29.4% were African American, and 71.4% were diabetic. The final random forest model provided C-statistics (95% confidence intervals) of 0.7185 (0.6994–0.7377), 0.7446 (0.7346–0.7546), 0.7504 (0.7425–0.7583), and 0.7488 (0.7421–0.7554) for predicting risk of death within the 4 different time windows. The models showed good internal validity and replicated well in patients with various demographic and clinical characteristics and provided similar or better performance compared with other ML algorithms. Results may not be generalizable to non-veterans. Use of predictors available in electronic medical records has limited the assessment of number of predictors. Conclusion: We implemented and ML-based method to accurately predict short-term postdialysis mortality in patients with incident ESRD. Our models could aid patients and clinicians in better decision making about the best course of action in patients approaching ESRD.

AB - Introduction: Given the high mortality rate within the first year of dialysis initiation, an accurate estimation of postdialysis mortality could help patients and clinicians in decision making about initiation of dialysis. We aimed to use machine learning (ML) by incorporating complex information from electronic health records to predict patients at risk for postdialysis short-term mortality. Methods: This study was carried out on a contemporary cohort of 27,615 US veterans with incident end-stage renal disease (ESRD). We implemented a random forest method on 49 variables obtained before dialysis transition to predict outcomes of 30-, 90-, 180-, and 365-day all-cause mortality after dialysis initiation. Results: The mean (±SD) age of our cohort was 68.7 ± 11.2 years, 98.1% of patients were men, 29.4% were African American, and 71.4% were diabetic. The final random forest model provided C-statistics (95% confidence intervals) of 0.7185 (0.6994–0.7377), 0.7446 (0.7346–0.7546), 0.7504 (0.7425–0.7583), and 0.7488 (0.7421–0.7554) for predicting risk of death within the 4 different time windows. The models showed good internal validity and replicated well in patients with various demographic and clinical characteristics and provided similar or better performance compared with other ML algorithms. Results may not be generalizable to non-veterans. Use of predictors available in electronic medical records has limited the assessment of number of predictors. Conclusion: We implemented and ML-based method to accurately predict short-term postdialysis mortality in patients with incident ESRD. Our models could aid patients and clinicians in better decision making about the best course of action in patients approaching ESRD.

UR - http://www.scopus.com/inward/record.url?scp=85070215161&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85070215161&partnerID=8YFLogxK

U2 - 10.1016/j.ekir.2019.06.009

DO - 10.1016/j.ekir.2019.06.009

M3 - Article

VL - 4

SP - 1219

EP - 1229

JO - Kidney International Reports

JF - Kidney International Reports

SN - 2468-0249

IS - 9

ER -