A survey of recent methods for addressing AI fairness and bias in biomedicine

Keyphrases

• Biomedicine 100%
• Artificial Intelligence 100%
• Data Perturbation 66%
• Artificial Intelligence Models 66%
• Debiasing Methods 66%
• Adversarial Learning 66%
• Clinical Practice 33%
• Clinical Setting 33%
• Computer Vision 33%
• Algorithmic Approach 33%
• Low Quality Data 33%
• Cost Reduction 33%
• Literature Search 33%
• Model-driven Development 33%
• Diagnostic Accuracy 33%
• Potential Bias 33%
• Model Making 33%
• PubMed 33%
• Data Augmentation 33%
• Irrelevant Features 33%
• General Domain 33%
• Loss-based 33%
• Sampling Bias 33%
• Artificial Intelligence Applications 33%
• Artificial Intelligence Systems 33%
• Data Sampling 33%
• Social Inequity 33%
• Multiple Combinations 33%
• Biomedical Domain 33%
• Unsupervised Representation Learning 33%
• ACM Digital Library 33%
• Surgical Decision Making 33%
• Learning Loss 33%
• Disentangled Representation Learning 33%
• Federated Learning 33%
• Learning Disentangled Representations 33%
• Clinical Natural Language Processing 33%
• Data Reweighting 33%
• IEEE Xplore 33%
• Distributional Method 33%
• Loose Constraints 33%

Medicine and Dentistry

• Intelligence 100%
• Health Care Cost 50%
• Diagnostic Accuracy 50%
• Decision Making 50%
• Weakness 50%
• Language Processing 50%

Pharmacology, Toxicology and Pharmaceutical Science

• Weakness 100%