A novel generative multi-task representation learning approach for predicting postoperative complications in cardiac surgery patients

Keyphrases

• Postoperative Complications 100%
• Cardiac Surgery Patients 100%
• Learning Approaches 100%
• Multi-Task Representation Learning 100%
• Variational Autoencoder 66%
• Patient Risk 50%
• Cardiac Surgery 33%
• AUPRC 33%
• Representation Learning 33%
• Retrospective Cohort Study 16%
• Blood Transfusion 16%
• Performance Prediction 16%
• Pulmonary Embolism 16%
• Deep Vein Thrombosis 16%
• Early Detection 16%
• Atrial Fibrillation 16%
• Cardiac Arrest 16%
• Acute Kidney Injury 16%
• Risk Profile 16%
• Patient Prognosis 16%
• Surgical Complications 16%
• Machine Learning 16%
• Data Prediction 16%
• Cardiac Events 16%
• Electronic Health Records 16%
• Surgical Patients 16%
• Risk Mitigation 16%
• Key Risk Factors 16%
• Superior Performance 16%
• Positive Events 16%
• Area under the Receiver Operating Characteristic (AUROC) 16%
• Under-five 16%
• Machine Learning Models 16%
• Learning Model 16%
• 5-fold Cross Validation 16%
• Year 2018 16%
• Data Complexity 16%
• Generative Models 16%
• Learning Framework 16%
• Discriminative Performance 16%
• Cross-cohort 16%
• Model Interpretation 16%
• Cohort Size 16%
• Cross-task 16%
• Machine Learning Solutions 16%
• Integrated Gradients 16%
• Best Alternative 16%

Medicine and Dentistry

• Heart Surgery 100%
• Postoperative Complication 100%
• Blood Transfusion 14%
• Retrospective Cohort Study 14%
• Pulmonary Embolism 14%
• Deep Vein Thrombosis 14%
• Atrial Fibrillation 14%
• Heart Arrest 14%
• Acute Kidney Injury 14%
• Cross-Validation 14%
• Electronic Health Record 14%