Algorithm Fairness in AI for Medicine and Healthcare
arXiv:2110.00603
Abstract
In the current development and deployment of many artificial intelligence (AI) systems in healthcare, algorithm fairness is a challenging problem in delivering equitable care. Recent evaluation of AI models stratified across race sub-populations have revealed inequalities in how patients are diagnosed, given treatments, and billed for healthcare costs. In this perspective article, we summarize the intersectional field of fairness in machine learning through the context of current issues in healthcare, outline how algorithmic biases (e.g. - image acquisition, genetic variation, intra-observer labeling variability) arise in current clinical workflows and their resulting healthcare disparities. Lastly, we also review emerging technology for mitigating bias via federated learning, disentanglement, and model explainability, and their role in AI-SaMD development.
References in corpus (20)
- Fairness in Machine Learning
- On Fairness and Calibration
- Differential Privacy-enabled Federated Learning for Sensitive Health Data
- Identifying and Correcting Label Bias in Machine Learning
- DALSA: Domain Adaptation for Supervised Learning From Sparsely Annotated MR Images
- Private Federated Learning with Domain Adaptation
- Iterative Orthogonal Feature Projection for Diagnosing Bias in Black-Box Models
- Federated and Differentially Private Learning for Electronic Health Records
- Mitigating Bias in Federated Learning
- How to be Fair and Diverse?
- Estimating Skin Tone and Effects on Classification Performance in Dermatology Datasets
- How does this interaction affect me? Interpretable attribution for feature interactions
- Improved Adversarial Learning for Fair Classification
- Assessing the validity of saliency maps for abnormality localization in medical imaging
- Fair Machine Learning Under Partial Compliance
- AI Fairness via Domain Adaptation
- FedDis: Disentangled Federated Learning for Unsupervised Brain Pathology Segmentation
- On Disentanglement in Gaussian Process Variational Autoencoders
- Federated Multi-Target Domain Adaptation
- Deep interpretability for GWAS