Dataset Information

Machine learning and deep learning to predict mortality in patients with spontaneous coronary artery dissection.

ABSTRACT: Machine learning (ML) and deep learning (DL) can successfully predict high prevalence events in very large databases (big data), but the value of this methodology for risk prediction in smaller cohorts with uncommon diseases and infrequent events is uncertain. The clinical course of spontaneous coronary artery dissection (SCAD) is variable, and no reliable methods are available to predict mortality. Based on the hypothesis that machine learning (ML) and deep learning (DL) techniques could enhance the identification of patients at risk, we applied a deep neural network to information available in electronic health records (EHR) to predict in-hospital mortality in patients with SCAD. We extracted patient data from the EHR of an extensive urban health system and applied several ML and DL models using candidate clinical variables potentially associated with mortality. We partitioned the data into training and evaluation sets with cross-validation. We estimated model performance based on the area under the receiver-operator characteristics curve (AUC) and balanced accuracy. As sensitivity analyses, we examined results limited to cases with complete clinical information available. We identified 375 SCAD patients of which mortality during the index hospitalization was 11.5%. The best-performing DL algorithm identified in-hospital mortality with AUC 0.98 (95% CI 0.97-0.99), compared to other ML models (P < 0.0001). For prediction of mortality using ML models in patients with SCAD, the AUC ranged from 0.50 with the random forest method (95% CI 0.41-0.58) to 0.95 with the AdaBoost model (95% CI 0.93-0.96), with intermediate performance using logistic regression, decision tree, support vector machine, K-nearest neighbors, and extreme gradient boosting methods. A deep neural network model was associated with higher predictive accuracy and discriminative power than logistic regression or ML models for identification of patients with ACS due to SCAD prone to early mortality.

SUBMITTER: Krittanawong C

PROVIDER: S-EPMC8076284 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:ImportanceSpontaneous coronary artery dissection (SCAD) is an increasingly recognized nonatherosclerotic cause of acute myocardial infarction enriched among individuals with early-onset myocardial infarction but is of unclear etiology.ObjectiveTo assess which genes contribute to the development of SCAD.Design, setting, and participantsTo prioritize genes influencing risk for SCAD, whole-exome sequencing was performed among individuals with SCAD in the discovery and replication cohorts from a tertiary care hospital outpatient specialty clinic, and gene set enrichment analyses were also performed for disruptive coding variants. All patients were sequentially enrolled beginning July 2013. Aggregate prevalence of rare disruptive variants for prioritized gene sets was compared between individuals with SCAD with population-based controls comprising 46 468 UK Biobank participants with whole-exome sequencing. Complementary mice models were used for in vivo validation. Analysis took place between June 2020 and January 2021.Main outcomes and measuresThe frequency and identity of rare genetic variants in individuals with SCAD.ResultsOf 130 patients, 109 (83.8%) were female (26 of 32 [81.2%] in the discovery cohort and 83 of 98 [84.7%] in the replication cohort) with mean (SD) age at first SCAD event of 48.41 (8.76) years in the discovery cohort and 47.74 (10.09) years in the replication cohort. Across all patients with SCAD, rare disruptive variants were found within 10 collagen genes (COL3A1, COL5A1, COL4A1, COL6A1, COL5A2, COL12A1, COL4A5, COL1A1, COL1A2, and COL27A1) were 17-fold (P = 1.5 × 10-9) enriched among individuals with SCAD compared with a background of 2506 constrained genes expressed in coronary artery. Furthermore, compared with individuals from the UK Biobank, individuals with SCAD were 1.75-fold (P = .04) more likely to carry disruptive rare variants within fibrillar collagen genes. Complementary mice models haploinsufficient for Col3a1 or Col5a1, the 2 most common collagen gene variants identified in SCAD cases, demonstrated increased risk of arterial dissection and increased size of arterial diameters especially in female mice, with resulting changes in collagen fibril organization and diameter.Conclusions and relevanceUnbiased gene discovery in patients with SCAD with independent human and murine validation highlights the role of the extracellular matrix dysfunction in SCAD.

Project description:ImportanceSpontaneous coronary artery dissection (SCAD) is a notable cause of acute coronary syndrome in women of childbearing age.ObjectiveTo test the hypothesis that pregnancy after SCAD is associated with recurrent SCAD.Design, setting, and participantsThree study designs were implemented: a case series of women with pregnancy after SCAD; a nested case-control study comparing patients with recurrent SCAD to matched controls without recurrent SCAD; and a cohort study. Women with SCAD who were of childbearing potential and enrolled into the Mayo Clinic SCAD Registry from August 30, 2011, to April 4, 2019, were included in the study. Patients with coronary dissections associated with iatrogenesis, trauma, or atherosclerosis were not enrolled.ExposuresPregnancy after SCAD.Main outcomes and measuresThe primary outcome was SCAD recurrence, defined as an acute coronary syndrome or cardiac arrest due to new SCAD. Other demographic measures collected included age, year of SCAD occurrence, and comorbidities.ResultsThe cohort included 636 women of childbearing potential. Twenty-three of those women had a total of 32 pregnancies after SCAD. The median (interquartile range) age of women with pregnancy after SCAD was 38 years (34-40 years), and 20 (87%) were White. In the nested case-control study, 92 cases of recurrent SCAD were matched to 158 controls. There was no significant difference in exposure to subsequent pregnancies in the women with recurrent SCAD as compared with matched controls (2 of 92 [2%] vs 13 of 158 [8%]; P = .06). In the overall cohort of 636 patients, recurrent SCAD was present in 122 patients with a Kaplan-Meier 5-year SCAD recurrence estimate of 14.8%. The Cox analysis showed no significant association between subsequent pregnancy and SCAD recurrence with a nonsignificant hazard ratio of 0.38 (95% CI, 0.09-1.6) when controlling for age at first SCAD, year of first SCAD, and fibromuscular dysplasia.Conclusions and relevanceThis study found that most women tolerated pregnancy and lactation after SCAD without evidence for increased risk of SCAD recurrence when compared with women with a history of SCAD who did not experience pregnancy. Although this study is reassuring and indicates complex contributors to SCAD recurrence, the results need to be interpreted prudently because of study selection bias and the small total number of women who became pregnant after SCAD. The notable hemodynamic changes that occur with pregnancy and severe presentation of pregnancy-associated SCAD are reasons for concern when considering pregnancy after SCAD.

Dataset Information

Machine learning and deep learning to predict mortality in patients with spontaneous coronary artery dissection.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets