Background The current ability to predict readmissions in patients with heart failure is modest at best. It is unclear whether machine learning (ML) techniques that address higher dimensional, non-linear relationships among variables would enhance prediction. We sought to compare the effectiveness of several ML algorithms for predicting readmissions. Methods and Results Using data from the Telemonitoring to Improve Heart Failure Outcomes trial, we compared the effectiveness of Random Forests (RF), Boosting, RF combined hierarchically with Support Vector Machines (SVM) or Logistic Regression (LR) and Poisson Regression against traditional LR to predict 30-day and 180-day all-cause and heart fauilre-only readmissions. We randomly selected 50% of patients for a derivation set and the remaining patients comprised a validation set, repeated 100 times. We compared c-statistics for discrimination and distributions of observed outcomes in risk deciles for predictive range. In 30-day all-cause readmission prediction, the best performing ML model, RF, provided a 17.8% improvement over LR (mean c-statistics 0.628 and 0.533, respectively). For readmissions due to heart failure, Boosting improved the c-statistic by 24.9% over LR (mean c-statistic 0.678 and 0.543, respectively). For 30-day all cause readmission, the observed readmission rates in the lowest and highest deciles of predicted risk with RF (7.8% and 26.2%, respectively) showed a much wider separation than LR (14.2% and 16.4%, respectively). Conclusions ML methods improved the prediction of readmission after hospitalization for heart failure compared with LR and provided the greatest predictive range in observed readmission rates.

Attrition rates are high despite mandated work hour reductions; 1 in 5 GS categorical residents resigns, and most pursue nonsurgical careers. Demographic factors, aside from postgraduate year do not appear predictive. Residents are at risk for attrition early in training and during research, and this could afford educators a target for intervention.

Aims Young women with acute myocardial infarction (AMI) have a higher risk of adverse outcomes than men. However, it is unclear how young women with AMI are different from young men across a spectrum of characteristics. We sought to compare young women and men at the time of AMI on 6 domains of demographic and clinical factors to determine whether they have distinct profiles. Methods and Results Using data from VIRGO, a prospective cohort study of women and men aged ≤55 years hospitalized for AMI(N=3,501) in the US and Spain, we evaluated sex differences in demographics, healthcare access, cardiovascular risk and psychosocial factors, symptoms and pre-hospital delay, clinical presentation, and hospital management for AMI. The study sample included 2,349(67%) women and 1,152(33%) men with mean age 47 years. Young women with AMI had higher rates of cardiovascular risk factors and comorbidities than men, including diabetes, congestive heart failure, chronic obstructive pulmonary disease, renal failure, and morbid obesity. They also exhibited higher levels of depression and stress, poorer physical and mental health status, and lower quality of life at baseline. Women had more delays in presentation and presented with higher clinical risk scores, on average, than men; however, men presented with higher levels of cardiac biomarkers and more classic electrocardiogram findings. Women were less likely to undergo revascularization procedures during hospitalization, and women with STEMI were less likely to receive timely primary reperfusion. Conclusions Young women with AMI represent a distinct, higher-risk population that is different from young men.

Background Studies of sex differences in long-term mortality after acute myocardial infarction (AMI) have reported mixed results. A systematic review is needed to characterize what is known about sex differences in long-term outcomes and to define gaps in knowledge. Methods and Results We searched the Medline database from 1966 to December 2012 to identify all studies that provided sex-based comparisons of mortality after AMI. Only studies with at least five years of follow-up were reviewed. Of the 1,877 identified abstracts, 52 studies met inclusion criteria, of which 39 were included in this review. Most studies included less than one-third women. There was significant heterogeneity across studies in patient populations, methodology, and risk adjustment, which produced substantial variability in risk estimates. In general, most studies reported higher unadjusted mortality for women compared with men at both 5 and 10 years after AMI; however, many of the differences in mortality became attenuated after adjustment for age. Multivariable models varied between studies; however, most reported a further reduction in sex differences after adjustment for covariates other than age. Few studies examined sex-by-age interactions; however several studies reported interactions between sex and treatment, whereby women have similar mortality risk as men after revascularization. Conclusions Sex differences in long-term mortality after AMI are largely explained by differences in age, comorbidities, and treatment utilization between women and men. Future research should aim to clarify how these differences in risk factors and presentation contribute to the sex gap in mortality.