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 (39)
- Communication-Efficient Learning of Deep Networks from Decentralized Data
- Axiomatic Attribution for Deep Networks
- Predicting Cardiovascular Risk Factors from Retinal Fundus Photographs using Deep Learning
- Federated Learning with Non-IID Data
- Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations
- Disentangling by Factorising
- Federated Optimization in Heterogeneous Networks
- A Benchmark for Interpretability Methods in Deep Neural Networks
- Large Scale Adversarial Representation Learning
- Fairness in Machine Learning
- On Fairness and Calibration
- Disentangled Representation Learning in Cardiac Image Analysis
- Differential Privacy-enabled Federated Learning for Sensitive Health Data
- Federated Adversarial Domain Adaptation
- Identifying and Correcting Label Bias in Machine Learning
- Learning Deep Disentangled Embeddings with the F-Statistic Loss
- Does Distributionally Robust Supervised Learning Give Robust Classifiers?
- DALSA: Domain Adaptation for Supervised Learning From Sparsely Annotated MR Images
- Attention-Gated Networks for Improving Ultrasound Scan Plane Detection
- Private Federated Learning with Domain Adaptation
- Iterative Orthogonal Feature Projection for Diagnosing Bias in Black-Box Models
- Calibration for the (Computationally-Identifiable) Masses
- Federated and Differentially Private Learning for Electronic Health Records
- Are All Training Examples Created Equal? An Empirical Study
- Fairness for Robust Log Loss Classification
- 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
- The implicit fairness criterion of unconstrained learning
- Improved Adversarial Learning for Fair Classification
- Algorithms for Fairness in Sequential Decision Making
- Assessing the validity of saliency maps for abnormality localization in medical imaging
- AI Fairness via Domain Adaptation
- Fair Machine Learning Under Partial Compliance
- FedDis: Disentangled Federated Learning for Unsupervised Brain Pathology Segmentation
- Federated Multi-Target Domain Adaptation
- On Disentanglement in Gaussian Process Variational Autoencoders
- Deep interpretability for GWAS