Giving Doctors an AI-Powered Head Start on Skin Cancer Detection of melanoma and other skin diseases is about to become faster and more accurate thanks to a new AI tool known as PanDerm. Developed by an international team of researchers led by Monash University, this innovative model is designed to assist doctors in real-world dermatological practice by analyzing multiple types of imaging simultaneously, including close-up photos, dermoscopic images, pathology slides, and total body photographs. The study, published in Nature Medicine, demonstrates the significant impact PanDerm could have in enhancing skin cancer diagnosis accuracy. When used by doctors, the model improved their diagnostic accuracy by 11%, and for non-dermatologist healthcare professionals, the improvement was even more pronounced, at 16.5%. One of the standout features of PanDerm is its ability to detect concerning lesions at an early stage, often before they are noticed by clinicians. This early detection capability is crucial for better treatment outcomes and patient management. The development of PanDerm was spearheaded by AI and computer vision expert Associate Professor Zongyuan Ge from Monash University’s Faculty of Information Technology. Existing AI models for dermatology have been limited in their scope, typically focusing on isolated tasks like diagnosing skin cancer from dermoscopic images. These models have struggled with integrating and processing various data types and imaging methods, which has reduced their utility in diverse clinical settings. PanDerm, however, was purposefully designed to work alongside clinicians, providing them with comprehensive, probabilistic assessments that enhance diagnostic confidence and decision-making. Ph.D. student Siyuan Yan, from Monash University’s Faculty of Engineering, emphasized that the multimodal approach—training the AI on diverse data from different imaging techniques—is the key to PanDerm's effectiveness. By synthesizing information from multiple visual sources, PanDerm can offer a more holistic analysis of skin conditions. This makes it uniquely capable of supporting a wide array of clinical tasks, from skin cancer screening to predicting the likelihood of cancer recurrence or spread, skin type assessment, mole counting, tracking lesion changes, and lesion segmentation. The importance of early detection in managing skin conditions cannot be overstated. According to first author and Ph.D. student Siyuan Yan, skin conditions now affect 70% of the global population. Early detection can significantly improve treatment outcomes, making tools like PanDerm potentially life-saving. Professor Victoria Mar, Director of the Victorian Melanoma Service at Alfred Health, noted that PanDerm's assistance in detecting subtle changes in lesions over time could lead to earlier diagnosis and more consistent monitoring for high-risk patients. Professor H. Peter Soyer, Director of the University of Queensland Dermatology Research Center, highlighted the practical benefits of PanDerm, especially in primary care settings where access to specialized dermatologists is often limited. The model's performance is robust, even when trained on a small amount of labeled data, which is a critical advantage in diverse medical environments. This characteristic makes PanDerm particularly valuable in resource-limited settings, such as rural and regional health care spaces, where standard annotated data may be scarce. Senior co-author Professor Harald Kittler from the Medical University of Vienna's Department of Dermatology underscored the significance of global collaboration and diverse clinical data in the development of AI tools. PanDerm's ability to support diagnosis across different health care systems, including in Europe, marks a notable advancement in making dermatological expertise more accessible and consistent worldwide. Despite these promising results, PanDerm is still in the evaluation phase and not yet widely implemented in health care settings. The researchers are committed to developing more comprehensive evaluation frameworks that cover a broader range of dermatological conditions and clinical contexts. They plan to establish standardized protocols for cross-demographic assessments and thoroughly investigate the model’s performance in varied real-world clinical settings. Ensuring equitable performance across different patient populations and health care environments remains a top priority. Industry Insights and Company Profiles Industry experts have praised the potential of PanDerm, highlighting its capacity to bridge the gap between technological innovation and clinical practice. The collaborative effort behind PanDerm, involving 11 institutions across multiple countries, underscores the growing trend in leveraging global data for AI development in healthcare. Monash University, known for its cutting-edge research in information technology and engineering, has again demonstrated its commitment to interdisciplinary solutions. The University of Queensland and the Medical University of Vienna have also played pivotal roles, contributing their extensive experience in dermatology and AI applications in medicine. These institutions’ collaboration showcases the potential for AI to revolutionize diagnostics and improve patient care, especially in areas where specialist resources are limited.