Equity in skin typing: why it is time to replace the Fitzpatrick scale

Disparities in dermatology AI performance on a diverse, curated clinical image set

“…While this is not comprehensive of all the potential confounders that could exist when comparing FST I–II and V–VI, our fine-tuning experiments show that the gap in performance can be closed by using diverse data. Although FST is used most commonly for labeling skin tones for images used in AI studies, this scale has limitations and does not capture the full diversity of human skin tones ( 26 ).…”

The essential role of dermatology publications in enhancing professional diversity, equity, and inclusion

“…The data collected should be of the highest possible quality. For that reason: Electronic health records should contain prompts to ensure recording of skin color and diversity of gender identification and racial and ethnic background, and the detailed reporting of these data should be in accepted articles. Drug companies, clinical research organizations, and institutions should ensure that clinical trial participants and investigators are inclusive and diverse. “Race and ethnicity” should be used in journal articles to denote the inexactness of concepts that are culturally constructed but may have implications for medical predisposition. Studies in Africa need to be encouraged because, for example, African American and Afro‐British research participants are not the same genetically and socioeconomically as African participants in Africa. Studies of skin and skin‐related diseases in indigenous populations throughout the world should be encouraged. Research targeted at conditions commonly seen in certain skin colors (eg, dark skin) should be prioritized, for example, keloids, pigmentation disorders, and tropical diseases. Research targeted at conditions commonly seen in the LGBTQ community and in economically underprivileged communities should be supported and encouraged. It is time to revisit and improve the Fitzpatrick skin color classification to denote more precisely the full range of skin colors and their variations in disease seen worldwide 2 …”

“…and socioeconomically as African participants in Africa.• Studies of skin and skin-related diseases in indigenous populations throughout the world should be encouraged.• Research targeted at conditions commonly seen in certain skin colors (eg, dark skin) should be prioritized, for example, keloids, pigmentation disorders, and tropical diseases.• Research targeted at conditions commonly seen in the LGBTQ community and in economically underprivileged communities should be supported and encouraged.• It is time to revisit and improve the Fitzpatrick skin color classification to denote more precisely the full range of skin colors and their variations in disease seen worldwide 2. While the above list is not exhaustive, and precise methodology must be developed, the intention is to drive change and support ef-forts to improve DEI in dermatology and other specialties.…”

Disparities in Dermatology AI: Assessments Using Diverse Clinical Images

“…This iteration of the DDI dataset was meant to capture the effects of skin tone on performance, thus images from FST I-II and FST V-VI were enriched. Additionally, though FST is used most commonly for labeling skin tones for images used in AI studies, this scale has limitations and does not capture the full diversity of human skin tones (Okoji et al, 2021). Since most dermatology AI algorithms are not shared, we were limited in the number of algorithms we could test (Daneshjou et al, 2021)…”