Home Medizin Untersuchungen zeigen, dass aktuelle KI-Programme bei der Erkennung von Hautläsionen bei farbigen Menschen schlechter abschneiden

Untersuchungen zeigen, dass aktuelle KI-Programme bei der Erkennung von Hautläsionen bei farbigen Menschen schlechter abschneiden

von NFI Redaktion

Artificial intelligence (AI) programs are gaining popularity in dermatology for the assessment, diagnosis, and treatment of skin conditions. A recent study published in the International Journal of Dermatology analyzed all published studies from the past 10 years to evaluate current AI programs used for dermatological purposes, revealing significant shortcomings in their application on skin of color (SOC).

The researchers identified various challenges in applying this technology to SOC, mainly due to the underrepresentation of SOC in datasets and issues with image quality and standardization. The findings suggest that current AI programs perform less accurately in detecting skin lesions on SOC. Additionally, only 30% of the programs identified in this analysis had specific data on SOC.

The idea for this research project originated from the establishment of South Jersey Skin Talk by lead author Rebecca Florent, an organization dedicated to educating others about dermatologic conditions, with a focus on those affecting people of color. Inspired by the significant lack of images representing skin of color, Rebecca examined the role of AI in dermatology, particularly its impact on addressing the needs of different skin types.

Thu Minh Truong, PharmD, RPh, corresponding author, medical student at Rutgers New Jersey Medical School

Source:

Journal Reference:

Florent, R., et al. (2024). Artificial Intelligence in Dermatology: Progress and Challenges in Skin of Color. International Journal of Dermatology. doi.org/10.1111/ijd.17076.

Related Posts

Adblock Detected

Please support us by disabling your AdBlocker extension from your browsers for our website.