Normal view MARC view ISBD view

Predicting Incident Heart Failure in Women With Machine Learning: The Women's Health Initiative Cohort.

MedStar author(s):

0828-282X

Name of journal: The Canadian journal of cardiologyAbstract: BACKGROUND: Heart failure (HF) is a leading cause of cardiac morbidity among women, whose risk factors differ from those in men. We used machine learning approaches to develop risk prediction models for incident HF in a cohort of postmenopausal women from the Women's Health Initiative (WHI).CONCLUSIONS: Machine learning approaches can be used to develop HF risk prediction models that can have better discrimination compared to an established HF risk model, and may provide a basis for investigating novel HF predictors. Copyright (c) 2021. Published by Elsevier Inc.METHODS AND RESULTS: We used two machine learning methods, Least Absolute Shrinkage and Selection Operator (LASSO) and Classification and Regression Trees (CART), to perform variable selection on 1,227 baseline WHI variables for the primary outcome of incident HF. These variables were then used to construct separate Cox proportional hazard models, and we compared these results, using receiver operating characteristic (ROC) curve analysis, against a comparator model built using variables from the Atherosclerosis Risk in Communities (ARIC) HF prediction model. We analyzed 43,709 women who had 2,222 incident HF events; median follow-up was 14.3 years. LASSO selected 10 predictors and CART selected 11 predictors. The highest correlation between selected variables was 0.46. In addition to selecting well-established predictors such as age, myocardial infarction, and smoking, novel predictors included physical function, number of pregnancies, number of prior live births and age at menopause. In ROC analysis, the CART-derived model had the highest c-statistic of 0.83 (95% CI 0.81-0.85), followed by LASSO 0.82 (95% CI 0.81-0.84) and ARIC 0.73 (95% 0.70-0.76).All authors: Allison MA, Avram R, Blair RH, Breathett K, Casanova R, Foraker RE, Howard BV, Klein L, Nah G, Olgin JE, Parikh NI, Tison GHOriginally published: Canadian Journal of Cardiology. 2021 Aug 13Fiscal year: FY2022Fiscal year of original publication: FY2022Digital Object Identifier:

https://dx.doi.org/10.1016/j.cjca.2021.08.006

Date added to catalog: 2021-11-01

Holdings ( 1 )
Title notes ( 4 )

Holdings
Item type	Current library	Collection	Call number	Status	Date due	Barcode
Journal Article	MedStar Authors Catalog	Article	34400272	Available		34400272

BACKGROUND: Heart failure (HF) is a leading cause of cardiac morbidity among women, whose risk factors differ from those in men. We used machine learning approaches to develop risk prediction models for incident HF in a cohort of postmenopausal women from the Women's Health Initiative (WHI).

CONCLUSIONS: Machine learning approaches can be used to develop HF risk prediction models that can have better discrimination compared to an established HF risk model, and may provide a basis for investigating novel HF predictors. Copyright (c) 2021. Published by Elsevier Inc.

METHODS AND RESULTS: We used two machine learning methods, Least Absolute Shrinkage and Selection Operator (LASSO) and Classification and Regression Trees (CART), to perform variable selection on 1,227 baseline WHI variables for the primary outcome of incident HF. These variables were then used to construct separate Cox proportional hazard models, and we compared these results, using receiver operating characteristic (ROC) curve analysis, against a comparator model built using variables from the Atherosclerosis Risk in Communities (ARIC) HF prediction model. We analyzed 43,709 women who had 2,222 incident HF events; median follow-up was 14.3 years. LASSO selected 10 predictors and CART selected 11 predictors. The highest correlation between selected variables was 0.46. In addition to selecting well-established predictors such as age, myocardial infarction, and smoking, novel predictors included physical function, number of pregnancies, number of prior live births and age at menopause. In ROC analysis, the CART-derived model had the highest c-statistic of 0.83 (95% CI 0.81-0.85), followed by LASSO 0.82 (95% CI 0.81-0.84) and ARIC 0.73 (95% 0.70-0.76).

English