Please login or create an account. If you do not have access to this content, you will be shown a 30 second preview and licensing options.

  • Presentation

How Diagnostic AI Can Change the Landscape of Dermatology

Description

In a session exploring the impact of diagnostic AI on dermatology, the speaker emphasized the potential of AI to enhance efficiency in patient care, particularly in diagnosing and monitoring skin conditions. They shared personal experiences, illustrating the time-consuming nature of tasks like obtaining medication approvals, and expressed a desire to leverage AI for more meaningful clinical work, such as biopsies and patient counseling. The speaker highlighted concerns regarding the limitations of current AI tools, particularly around data sets that predominantly feature images of lighter-skinned individuals, which leads to inadequate performance on diverse populations. They discussed various AI applications, noting that while some have shown promise in accurately diagnosing conditions such as melanoma, many are still not adequately validated and may prioritize business models over patient care. The importance of using AI to help primary care providers with diagnostic confidence was also mentioned, along with the need for caution regarding AI's future role in prescribing medications without proper oversight. Overall, the speaker advocated for a thoughtful and critical approach to integrating AI into dermatological practice, stressing the need for ongoing evaluation of its efficacy and safety.

View more

Conclusions

  • Diagnostic AI can enhance the diagnosis and monitoring of skin conditions.
  • AI tools may outperform primary care providers and even some dermatologists in diagnosing skin issues.
  • Current AI models generally require training on diverse datasets to avoid biases and inaccuracies.
  • Commercial AI applications in dermatology often lack adequate validation, leading to unreliable results.
  • AI can improve efficiency and access to care in dermatology, especially in underserved areas.
  • Patients increasingly rely on AI tools out of necessity due to high healthcare costs and limited access to dermatologists.
  • AI should be used to support, not replace, the clinical judgment of dermatologists and healthcare providers.
  • The accuracy of an AI tool was shown to be clinically significant and could meaningfully impact patient care in identifying melanomas.
  • Börve, A. (2019). Google dermatology AI detects 26 Skin Conditions as Accurately as Dermatologists. First Derm. Retrieved from [link]
  • Buolamwini, J. (2017). Gender Shades: Intersectional Phenotypic and Demographic Evaluation of Face Datasets and Gender Classifiers. MIT Master's Thesis.
  • Papachristou, P., Söderholm, M., Pallon, J., Taloyan, M., & Polesie, S. (2024). Evaluation of an artificial intelligence-based decision support for the detection of cutaneous melanoma in primary care: a prospective real-life clinical trial. British Journal of Dermatology, 191(1), 125-133. https://doi.org/10.1093/bjd/ljae021
  • Dulmage, B., Tegtmeyer, K., Zhang, M. Z., & Colavincenzo, M. (2021). A Point-of-Care, Real-Time Artificial Intelligence System to Support Clinician Diagnosis of a Wide Range of Skin Diseases. Journal of Investigative Dermatology, 141(5), 1230-1235.