Fairness of Classifiers Across Skin Tones in Dermatology

Kinyanjui, Newton Mwai; Odonga, Timothy; Cintas, Celia; Codella, Noel; Panda, Rameswar; Sattigeri, Prasanna; Varshney, Kush R.

doi:10.1007/978-3-030-59725-2_31

Cited by 66 publications

(44 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Skin tone was objectively measured from the participants’ cheek skin using a Pantone Capsure color matcher colorimeter (X-Rite, Grand Rapids, MI). Evidence suggests that darker skin tone is frequently under-represented in medical datasets, 22 and that medical devices using optical sensors may be less accurate in those individuals. 23–25 Therefore, the darkest skin-tone subgroup was intentionally oversampled to ensure the algorithm’s unbiased performance over various skin tones.…”

Section: Methodsmentioning

confidence: 99%

Prospective validation of smartphone-based heart rate and respiratory rate measurement algorithms

Bae

Borac

Emre

et al. 2021

Preprint

View full text Add to dashboard Cite

Measuring vital signs plays a key role in both patient care and wellness, but can be challenging outside of medical settings due to the lack of specialized equipment. In this study, we prospectively evaluated smartphone camera-based techniques for measuring heart rate (HR) and respiratory rate (RR) for consumer wellness use. HR was measured by placing the finger over the rear-facing camera, while RR was measured via a video of the participants sitting still in front of the front-facing camera. In the HR study of 95 participants (with a protocol that included both measurements at rest and post exercise), the mean absolute percent error (MAPE) ± standard deviation of the measurement was 1.6% ± 4.3%, which was significantly lower than the pre-specified goal of 5%. No significant differences in the MAPE were present across colorimeter-measured skin-tone subgroups: 1.8% ± 4.5% for very light to intermediate, 1.3% ± 3.3% for tan and brown, and 1.8% ± 4.9% for dark. In the RR study of 50 participants, the mean absolute error (MAE) was 0.78 ± 0.61 breaths/min, which was significantly lower than the pre-specified goal of 3 breath/min. The MAE was low in both healthy participants (0.70 ± 0.67 breaths/min), and participants with chronic respiratory conditions (0.80 ± 0.60 breaths/min). Our results validate that smartphone camera-based techniques can accurately measure HR and RR across a range of pre-defined subgroups.

show abstract

Section: Methodsmentioning

confidence: 99%

Prospective validation of smartphone-based heart rate and respiratory rate measurement algorithms

Bae

Borac

Emre

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Then we utilize the recent Generalized Histogram Thresholding (GHT) [3] method for segmentation, which kept surrounding skin of the lesions only. We followed [16] transformed images from RGB-space to CIELab-space to calculate the ITA value of the non-lesion area. We set skins with ITA less than 45 as the dark skins.…”

Section: B Data Prepossessingmentioning

confidence: 99%

“…Existing chest X-ray classifiers were found exhibiting disparities in performance across subgroups distinguished by different phenotypes [22,22]. Similar to forms of discrimination linked to face recognition [27], darker skinned patients may be under-presented [16] in existing dermatology datasets: ISIC 2018 Challenge dataset [25,8]. A further study [1] demonstrated that the skin lesion algorithm could show variable performance relative to other under-represented subgroups.…”

Section: Introductionmentioning

confidence: 99%

Estimating and Improving Fairness with Adversarial Learning

Li,

Cui,

et al. 2021

Preprint

View full text Add to dashboard Cite

Fairness and accountability are two essential pillars for trustworthy Artificial Intelligence (AI) in healthcare. However, the existing AI model may be biased in its decision marking. To tackle this issue, we propose an adversarial multi-task training strategy to simultaneously mitigate and detect bias in the deep learning-based medical image analysis system. Specifically, we propose to add a discrimination module against bias and a critical module that predicts unfairness within the base classification model. We further impose an orthogonality regularization to force the two modules to be independent during training. Hence, we can keep these deep learning tasks distinct from one another, and avoid collapsing them into a singular point on the manifold. Through this adversarial training method, the data from the under-privileged group, which is vulnerable to bias because of attributes such as sex and skin tone, are transferred into a domain that is neutral relative to these attributes. Furthermore, the critical module can predict fairness scores for the data with unknown sensitive attributes. We evaluate our framework on a large-scale public-available skin lesion dataset under various fairness evaluation metrics. The experiments demonstrate the effectiveness of our proposed method for estimating and improving fairness in the deep learning-based medical image analysis system.

show abstract

“…Using gender-balanced datasets, a recent study found statistically significant disparities in performance on medical imaging-based diagnosis [11]. Darker skinned patients may be underrepresented [10,13] in existing dermatology datasets [20,3], similar to forms of prejudice associated to face recognition [22]. The effects of decision-making that is (partially) based on the values of biased qualities can be irrevocable or even lethal, especially in medical applications.…”

Section: Fairnessmentioning

confidence: 99%

On the Fairness of Swarm Learning in Skin Lesion Classification

Fan¹,

Wu²,

Li³

2021

Preprint

View full text Add to dashboard Cite

Fairness is essential for trustworthy Artificial Intelligence (AI) in healthcare. However, the existing AI model may be biased in its decision marking. The bias induced by data itself, such as collecting data in subgroups only, can be mitigated by including more diversified data. Distributed and collaborative learning is an approach to involve training models in massive, heterogeneous, and distributed data sources, also known as nodes. In this work, we target on examining the fairness issue in Swarm Learning (SL), a recent edge-computing based decentralized machine learning approach, which is designed for heterogeneous illnesses detection in precision medicine. SL has achieved high performance in clinical applications, but no attempt has been made to evaluate if SL can improve fairness. To address the problem, we present an empirical study by comparing the fairness among single (node) training, SL, centralized training. Specifically, we evaluate on large public available skin lesion dataset, which contains samples from various subgroups. The experiments demonstrate that SL does not exacerbate the fairness problem compared to centralized training and improves both performance and fairness compared to single training. However, there still exists biases in SL model and the implementation of SL is more complex than the alternative two strategies.

show abstract

Fairness of Classifiers Across Skin Tones in Dermatology

Cited by 66 publications

References 25 publications

Prospective validation of smartphone-based heart rate and respiratory rate measurement algorithms

Prospective validation of smartphone-based heart rate and respiratory rate measurement algorithms

Estimating and Improving Fairness with Adversarial Learning

On the Fairness of Swarm Learning in Skin Lesion Classification

Contact Info

Product

Resources

About