Marble: High-throughput phenotyping from electronic health records via sparse nonnegative tensor factorization

Keyphrases

• Electronic Health Records 100%
• High-throughput Phenotyping 100%
• Nonnegative Tensor Factorization 100%
• Calcium Carbonate 100%
• Baseline Characteristics 50%
• Electronic Health Record Data 50%
• Interaction Tensor 50%
• Patient Demographics 25%
• Large Segment 25%
• Publicly Available 25%
• Patient Management 25%
• Clinical Research 25%
• Predictive Power 25%
• Patient Data 25%
• Clinical Investigator 25%
• Data-driven Approach 25%
• Human Supervision 25%
• Heterogeneous Data Sources 25%
• Factorization Method 25%
• Zero Element 25%
• Medical Concepts 25%
• Clinical Database 25%
• Medical Experts 25%
• Decision-making Management 25%
• Data-driven Solutions 25%
• Intensive Supervision 25%
• Research Preferences 25%
• Phenotype Characteristic 25%

Computer Science

• High Throughput 100%
• Data Record 100%
• Predictive Power 50%
• Driven Approach 50%
• Common Characteristic 50%
• Nonzero Element 50%
• Clinical Researcher 50%
• Decision Making 50%