Comment on “Artificial Intelligence in melanoma research: a bibliometric analysis”

Dear Editor, Melanoma is a cancer that arises from epidermal melanocytes[1–3]. Ultraviolet radiation (UVR) exposure is intimately associated with its pathogenesis, and intermittent high-intensity UVR exposure is found to be the main risk factor for melanoma[4–6]. A new era in the medical industry was brought about by the development of artificial intelligence (AI) technology[7,8]. AI in melanoma research is a quickly emerging interdisciplinary subject that integrates molecular biology, computer science, and oncology to address important problems with diagnosis, therapy, and prognosis. The rapid expansion of this field’s literature suggests that a thorough examination of its intellectual structure and dynamics is crucial. Nevertheless, the use of bibliometrics in AI melanoma research has not been thoroughly examined. To fill this gap, Guo et al[9] conducted a bibliometric analysis. There were 1325 articles and 151 reviews in all (Fig. 1). After 2017, they discovered that the number of publications suddenly increased every year, reaching 260 in 2024. The most productive countries in terms of publications are the United States (337 publications) and China (292 publications). Germany is the top contributor in terms of authors and institutions. The keyword analysis revealed that AI-integrated immunotherapy and AI-assisted diagnostics were the two most important research hotspots. Seven of the top 10 cited papers are mostly focused on deep learning for image classification, segmentation, as well as diagnostic performance versus dermatologists, indicating a predominant focus on AI applications for melanoma diagnosis. After reading this paper, we want to share some opinions and limitations about the topic. Figure 1.: Flow chart of literature search and bibliometric analysis. Several included studies, especially from Asian countries, have mentioned that there were some difficulties when utilizing AI diagnosis, including communication with patients and AI technical factors. Based on our clinical experiences, it is important to strike a balance between providing patients with sufficient information and avoiding excessive technical jargon while explaining the AI diagnosis procedure. Clinicians should be ready to discuss how AI evaluates images, how it helps with diagnosis, and how important human oversight is to the ultimate decision-making process. This demonstrates how AI enhances clinical judgment rather than replaces it. A significant challenge arises when a clinician’s assessment of a melanoma image varies from the AI’s interpretation. In these situations, the dermatologist must carefully consider both points of view in order to identify the cause of the dispute. This might involve reexamining the lesion, asking a colleague for a second opinion, or analyzing other clinical data that AI is unable to access. Enhancing validation frameworks and data diversity, developing explainable AI approaches to foster clinical trust, optimizing immunotherapy response prediction through multi-omics integration, and bolstering cooperation between AI developers and clinicians to overcome implementation obstacles are the main areas of future research priorities. This study was conducted in compliance with the Transparency in the Reporting of Artificial Intelligence – the TITAN guideline[10].