Machine Learning and Health Care Disparities in Dermatology.

By A. Adamson, Avery Smith. 2018

493 citations recorded by the source corpus.

Subfields: Dermatology, Oncology, AI Ethics, Fairness & Society, Medical & Health AI

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Dermatology

1. James Y. Zou

In Oncology

1. John R. Smith
2. Michael Alexander Riegler
3. Gary S. Collins
4. Karel G. M. Moons
5. Noel C. F. Codella

In AI Ethics, Fairness & Society

1. James Y. Zou
2. Rogério Schmidt Feris
3. John R. Smith

In Medical & Health AI

1. Fei Wang
2. Rogério Schmidt Feris
3. Gary S. Collins
4. John R. Smith
5. Kate Saenko
6. Evangelia Christodoulou
7. Anita M. Preininger
8. B. van Ginneken

Citations in the corpus

• AI in Health: State of the Art, Challenges, and Future Directions (2019, background): “dark skins) may introduce disparities in the model’s decisions implicitly [27].”
• A new direction to promote the implementation of artificial intelligence in natural clinical settings (2019, background): “In addition, Adamson et al. 34 pointed out another issue to be addressed for the AI system: it should hold the inclusivity to handle different types of skin ( Category A ), i.e., skins with different colors.”
• Ensuring Fairness in Machine Learning to Advance Health Equity (2018, methodology): “For example, if advanced-stage melanoma is diagnosed more frequently in patients with dark skin than in other groups, might a skin cancer detection model fail to detect early-stage disease in patients with dark skin (47, 48)? During training and evaluation, researchers should measure any deviations ”
• A New Benchmark for Evaluation of Cross-Domain Few-Shot Learning (2019, background): “However, collecting sufficient amounts of data for certain classes may be impossible in practice: for example, in dermatology, there are a multitude of instances of rare diseases, or diseases that become rare for particular types of skin [28, 1, 14].”
• Metrics Reloaded: Recommendations for image analysis validation (2022, background): “Given that learning-based algorithms rely on the availability of historical datasets for training, there is a real risk that any existing biases in the data may be picked up and replicated or even exacerbated when an algorithm makes predictions 52,53 .”
• A Broader Study of Cross-Domain Few-Shot Learning (2019, background): “However, collecting sufficient amounts of data for certain classes may be impossible in practice: for example, in dermatology, there are a multitude of instances of rare diseases, or diseases that become rare for particular types of skin [48,1,25].”
• Shifting machine learning for healthcare from development to deployment and from models to data (2022, background): “…classifiers that were trained primarily on images of one skin tone decrease in performance when evaluated on images of different skin tones 173 ; in this case, the drop in performance could be attributed to variations in disease presentation that are not captured when certain patient…”
• Ethics and governance of trustworthy medical artificial intelligence (2023, background): “For example, an AI for melanoma diagnosis is difficult to apply to darker races because there are few medical images of darker races in the skin lesion database [29].”