Multi-step Transfer Learning in Natural Language Processing for the Health Domain

Keyphrases

• Verbal Autopsy 100%
• Autopsy Report 60%
• Learning Words 40%
• Biomedical Domain 40%
• ELMo 40%
• Patient Confidentiality 20%
• Adaptation Techniques 20%
• Patient Privacy 20%
• Standard English 20%
• Cause of Death Classification 20%
• Transformer Embeddings 20%
• Model Embedding 20%
• Multiscale Feature Learning 20%
• Two-component Domains 20%
• Confidentiality Policy 20%
• Non-clinicians 20%

Computer Science

• Embedding From Language Model 66%
• Clinical Data 33%
• Text Representation 33%