Dataset Information

Machine Learning Improves Risk Stratification After Acute Coronary Syndrome.

ABSTRACT: The accurate assessment of a patient's risk of adverse events remains a mainstay of clinical care. Commonly used risk metrics have been based on logistic regression models that incorporate aspects of the medical history, presenting signs and symptoms, and lab values. More sophisticated methods, such as Artificial Neural Networks (ANN), form an attractive platform to build risk metrics because they can easily incorporate disparate pieces of data, yielding classifiers with improved performance. Using two cohorts consisting of patients admitted with a non-ST-segment elevation acute coronary syndrome, we constructed an ANN that identifies patients at high risk of cardiovascular death (CVD). The ANN was trained and tested using patient subsets derived from a cohort containing 4395 patients (Area Under the Curve (AUC) 0.743) and validated on an independent holdout set containing 861 patients (AUC 0.767). The ANN 1-year Hazard Ratio for CVD was 3.72 (95% confidence interval 1.04-14.3) after adjusting for the TIMI Risk Score, left ventricular ejection fraction, and B-type natriuretic peptide. A unique feature of our approach is that it captures small changes in the ST segment over time that cannot be detected by visual inspection. These findings highlight the important role that ANNs can play in risk stratification.

SUBMITTER: Myers PD

PROVIDER: S-EPMC5627253 | biostudies-literature | 2017 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine Learning Improves Risk Stratification After Acute Coronary Syndrome.

Myers Paul D PD Scirica Benjamin M BM Stultz Collin M CM

Scientific reports 20171004 1

The accurate assessment of a patient's risk of adverse events remains a mainstay of clinical care. Commonly used risk metrics have been based on logistic regression models that incorporate aspects of the medical history, presenting signs and symptoms, and lab values. More sophisticated methods, such as Artificial Neural Networks (ANN), form an attractive platform to build risk metrics because they can easily incorporate disparate pieces of data, yielding classifiers with improved performance. Us ...[more]

PMID: 28978948

Similar Datasets

Project description:Background Cystatin C (Cys-C) is a marker of renal function that has shown prognostic value for cardiovascular risk stratification across different patient populations. The incremental value of Cys-C beyond established cardiac and renal biomarkers remains incompletely explored. Methods and Results SOLID - TIMI 52 (Stabilization of Plaques Using Darapladib-Thrombolysis in Myocardial Infarction 52; www.clinicaltrials.gov , NCT01000727) randomized patients ≤30 days post-acute coronary syndrome were treated with darapladib or placebo. The association between Cys-C and long-term risk (median follow-up 2.5 years) was assessed in 4965 individuals with adjustments made for clinical variables and other risk markers (eg, estimated glomerular filtration rate, high-sensitivity troponin I, brain-type natriuretic peptide, and fibroblast growth factor-23). The prespecified outcome of interest was cardiovascular death (CVD) or heart failure hospitalization. Cys-C was strongly correlated with creatinine ( r=0.60) and estimated glomerular filtration rate ( r=-0.68), moderately correlated with fibroblast growth factor-23 ( r=0.39), and weakly correlated with brain-type natriuretic peptide ( r=0.28) and high-sensitivity troponin I ( r=0.06) (all P<0.0001). After multivariate adjustment, increasing concentration of Cys-C (per SD of log-transformed Cys-C) was significantly associated with a 28% higher hazard of CVD or heart failure hospitalization (hazard ratio [ HR ] 1.28, 95% confidence interval [ CI ] 1.12-1.46, P<0.001), including CVD ( HR 1.24, 95% CI 1.04-1.47, P=0.01) and heart failure hospitalization ( HR 1.42, 95% CI 1.19-1.69, P<0.001). Cys-C was also associated with a higher hazard of CVD, myocardial infarction, or stroke ( HR 1.15, 95% CI 1.04-1.28, P<0.01), including myocardial infarction ( HR 1.17, 95% CI 1.02-1.33, P=0.02). The addition of Cys-C to a fully adjusted model without estimated glomerular filtration rate improved the C-statistic from 0.80 to 0.81 ( P=0.03) for CVD or heart failure hospitalization. In contrast, the addition of estimated glomerular filtration rate to a fully adjusted model without Cys-C failed to improve model discrimination ( P=0.17). Conclusions Cys-C is associated with the risk of adverse outcomes in patients after acute coronary syndrome. This relationship is independent of established and novel biomarkers of the cardiorenal axis.

Dataset Information

Machine Learning Improves Risk Stratification After Acute Coronary Syndrome.

Publications

Machine Learning Improves Risk Stratification After Acute Coronary Syndrome.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets