BackgroundAtrial fibrillation is associated with higher mortality. Identification of causes of death and contemporary risk factors for all‐cause mortality may guide interventions.Methods and ResultsIn the Rivaroxaban Once Daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF) study, patients with nonvalvular atrial fibrillation were randomized to rivaroxaban or dose‐adjusted warfarin. Cox proportional hazards regression with backward elimination identified factors at randomization that were independently associated with all‐cause mortality in the 14 171 participants in the intention‐to‐treat population. The median age was 73 years, and the mean CHADS 2 score was 3.5. Over 1.9 years of median follow‐up, 1214 (8.6%) patients died. Kaplan–Meier mortality rates were 4.2% at 1 year and 8.9% at 2 years. The majority of classified deaths (1081) were cardiovascular (72%), whereas only 6% were nonhemorrhagic stroke or systemic embolism. No significant difference in all‐cause mortality was observed between the rivaroxaban and warfarin arms (P=0.15). Heart failure (hazard ratio 1.51, 95% CI 1.33–1.70, P<0.0001) and age ≥75 years (hazard ratio 1.69, 95% CI 1.51–1.90, P<0.0001) were associated with higher all‐cause mortality. Multiple additional characteristics were independently associated with higher mortality, with decreasing creatinine clearance, chronic obstructive pulmonary disease, male sex, peripheral vascular disease, and diabetes being among the most strongly associated (model C‐index 0.677).ConclusionsIn a large population of patients anticoagulated for nonvalvular atrial fibrillation, ≈7 in 10 deaths were cardiovascular, whereas <1 in 10 deaths were caused by nonhemorrhagic stroke or systemic embolism. Optimal prevention and treatment of heart failure, renal impairment, chronic obstructive pulmonary disease, and diabetes may improve survival.Clinical Trial Registration URL: https://www.clinicaltrials.gov/. Unique identifier: NCT00403767.

This paper presents a study showing the benefits of the Effi-cientNet models compared with heavier Convolutional Neural Networks (CNNs) in the Document Classification task, essential problem in the digitalization process of institutions. We show in the RVL-CDIP dataset that we can improve previous results with a much lighter model and present its transfer learning capabilities on a smaller in-domain dataset such as Tobacco3482. Moreover, we present an ensemble pipeline which is able to boost solely image input by combining image model predictions with the ones generated by BERT model on extracted text by OCR. We also show that the batch size can be effectively increased without hindering its accuracy so that the training process can be sped up by parallelizing throughout multiple GPUs, decreasing the computational time needed. Lastly, we expose the training performance differences between PyTorch and Tensorflow Deep Learning frameworks.

BACKGROUND Guidelines recommend nonstatin lipid-lowering agents in patients at very high risk for major adverse cardiovascular events (MACE) if low-density lipoprotein cholesterol (LDL-C) remains ≥70 mg/dL on maximum tolerated statin treatment. It is uncertain if this approach benefits patients with LDL-C near 70 mg/dL. Lipoprotein(a) levels may influence residual risk. OBJECTIVES In a post hoc analysis of the ODYSSEY Outcomes (Evaluation of Cardiovascular Outcomes After an Acute Coronary Syndrome During Treatment With Alirocumab) trial, the authors evaluated the benefit of adding the proprotein subtilisin/kexin type 9 inhibitor alirocumab to optimized statin treatment in patients with LDL-C levels near 70 mg/dL. Effects were evaluated according to concurrent lipoprotein(a) levels. METHODS ODYSSEY Outcomes compared alirocumab with placebo in 18,924 patients with recent acute coronary syndromes receiving optimized statin treatment. In 4,351 patients (23.0%), screening or randomization LDL-C was <70 mg/dL (median 69.4 mg/dL; interquartile range: 64.3–74.0 mg/dL); in 14,573 patients (77.0%), both determinations were ≥70 mg/dL (median 94.0 mg/dL; interquartile range: 83.2–111.0 mg/dL). RESULTS In the lower LDL-C subgroup, MACE rates were 4.2 and 3.1 per 100 patient-years among placebo-treated patients with baseline lipoprotein(a) greater than or less than or equal to the median (13.7 mg/dL). Corresponding adjusted treatment hazard ratios were 0.68 (95% confidence interval [Cl]: 0.52–0.90) and 1.11 (95% Cl: 0.83–1.49), with treatment-lipoprotein(a) interaction on MACE ( P interaction = 0.017). In the higher LDL-C subgroup, MACE rates were 4.7 and 3.8 per 100 patient-years among placebo-treated patients with lipoprotein(a) >13.7 mg/dL or ≤13.7 mg/dL; corresponding adjusted treatment hazard ratios were 0.82 (95% Cl: 0.72–0.92) and 0.89 (95% Cl: 0.75–1.06), with P interaction = 0.43. CONCLUSIONS In patients with recent acute coronary syndromes and LDL-C near 70 mg/dL on optimized statin therapy, proprotein subtilisin/kexin type 9 inhibition provides incremental clinical benefit only when lipoprotein(a) concentration is at least mildly elevated. (ODYSSEY Outcomes: Evaluation of Cardiovascular Outcomes After an Acute Coronary Syndrome During Treatment With Alirocumab; NCT01663402 )

No abstract

No abstract