Single-Lesion Skin Cancer Risk Stratification Triage Pathway.

Importance: Access to timely dermatologic care remains a challenge, especially for patients with new skin lesions. Assessing the efficiency of new triage pathways may assist in better resource allocation and shorter time to care. Objective: To evaluate whether a rule-based triage system was associated with better skin cancer risk stratification of patients and reduced wait times. Design, Setting, and Participants: This retrospective quality improvement cohort study of patients referred to Stanford University dermatology clinics was conducted between November 2017 and January 2023. A rules-based triage system based on a priori-determined high-risk lesion characteristics was implemented. Exposures: Referral reasons and risk factors of patients provided by their primary care physicians. Main Outcomes and Measures: Biopsy results of patients (diagnosis of any skin cancer and melanoma) at their visit or within 6 months after the visit. Regression models were used to assess the association between risk factors at referral and (1) biopsy outcomes and (2) time to first visit, adjusting for sociodemographic factors. Results: Among 37 478 patients (mean [SD] age, 54 (18) years; 21 292 women [57%]), the rates of aggregate biopsy, malignant biopsy specimens, and melanoma were comparable across patients seen after (n = 12 302) and before (n = 25 176) the implementation of the new triage pathway. Patients seen through the lesion pathway had a higher risk of having malignant biopsy results (adjusted risk ratio [aRR], 1.6; 95% CI, 1.4-1.9) and melanoma (aRR, 2.0; 95% CI, 1.2-3.2) than those not seen through the pathway. Lesions that were concerning to referring clinicians for skin cancer were associated with an increased risk of skin cancer (all skin cancer: aRR, 2.8; 95% CI, 2.2-3.5; melanoma: aRR, 2.02; 95% CI, 1.1-3.7). Patients in the 3 high-risk lesion groups were seen faster in the new triage pathway (mean reduction, 26 days; 95% CI, 18-34 days). Conclusions and Relevance: In this study, a new automated, rules-based referral pathway was implemented that expedited care for patients with high-risk skin cancer. This reform may have contributed to improving patient stratification, reducing the time from referral to first encounter, and maintaining accuracy in identifying malignant lesions. The findings highlight the potential to optimize clinical resource allocation by better risk stratification of referred patients.

Human-AI interaction in skin cancer diagnosis: a systematic review and meta-analysis.

The development of diagnostic tools for skin cancer based on artificial intelligence (AI) is increasing rapidly and will likely soon be widely implemented in clinical use. Even though the performance of these algorithms is promising in theory, there is limited evidence on the impact of AI assistance on human diagnostic decisions. Therefore, the aim of this systematic review and meta-analysis was to study the effect of AI assistance on the accuracy of skin cancer diagnosis. We searched PubMed, Embase, IEE Xplore, Scopus and conference proceedings for articles from 1/1/2017 to 11/8/2022. We included studies comparing the performance of clinicians diagnosing at least one skin cancer with and without deep learning-based AI assistance. Summary estimates of sensitivity and specificity of diagnostic accuracy with versus without AI assistance were computed using a bivariate random effects model. We identified 2983 studies, of which ten were eligible for meta-analysis. For clinicians without AI assistance, pooled sensitivity was 74.8% (95% CI 68.6-80.1) and specificity was 81.5% (95% CI 73.9-87.3). For AI-assisted clinicians, the overall sensitivity was 81.1% (95% CI 74.4-86.5) and specificity was 86.1% (95% CI 79.2-90.9). AI benefitted medical professionals of all experience levels in subgroup analyses, with the largest improvement among non-dermatologists. No publication bias was detected, and sensitivity analysis revealed that the findings were robust. AI in the hands of clinicians has the potential to improve diagnostic accuracy in skin cancer diagnosis. Given that most studies were conducted in experimental settings, we encourage future studies to further investigate these potential benefits in real-life settings.

Transparent medical image AI via an image-text foundation model grounded in medical literature.

Building trustworthy and transparent image-based medical artificial intelligence (AI) systems requires the ability to interrogate data and models at all stages of the development pipeline, from training models to post-deployment monitoring. Ideally, the data and associated AI systems could be described using terms already familiar to physicians, but this requires medical datasets densely annotated with semantically meaningful concepts. In the present study, we present a foundation model approach, named MONET (medical concept retriever), which learns how to connect medical images with text and densely scores images on concept presence to enable important tasks in medical AI development and deployment such as data auditing, model auditing and model interpretation. Dermatology provides a demanding use case for the versatility of MONET, due to the heterogeneity in diseases, skin tones and imaging modalities. We trained MONET based on 105,550 dermatological images paired with natural language descriptions from a large collection of medical literature. MONET can accurately annotate concepts across dermatology images as verified by board-certified dermatologists, competitively with supervised models built on previously concept-annotated dermatology datasets of clinical images. We demonstrate how MONET enables AI transparency across the entire AI system development pipeline, from building inherently interpretable models to dataset and model auditing, including a case study dissecting the results of an AI clinical trial.

The Promises and Perils of Foundation Models in Dermatology.

Foundation models (FM), which are large-scale artificial intelligence (AI) models that can complete a range of tasks, represent a paradigm shift in AI. These versatile models encompass large language models, vision-language models, and multimodal models. Although these models are often trained for broad tasks, they have been applied either out of the box or after additional fine tuning to tasks in medicine, including dermatology. From addressing administrative tasks to answering dermatology questions, these models are poised to have an impact on dermatology care delivery. As FMs become more ubiquitous in health care, it is important for clinicians and dermatologists to have a basic understanding of how these models are developed, what they are capable of, and what pitfalls exist. In this paper, we present a comprehensive yet accessible overview of the current state of FMs and summarize their current applications in dermatology, highlight their limitations, and discuss future developments in the field.

Current State of Dermatology Mobile Applications With Artificial Intelligence Features.

Importance: With advancements in mobile technology and artificial intelligence (AI) methods, there has been a substantial surge in the availability of direct-to-consumer mobile applications (apps) claiming to aid in the assessment and management of diverse skin conditions. Despite widespread patient downloads, these apps exhibit limited evidence supporting their efficacy. Objective: To identify and characterize current English-language AI dermatology mobile apps available for download, focusing on aspects such as purpose, supporting evidence, regulatory status, clinician input, data privacy measures, and use of image data. Evidence Review: In this cross-sectional study, both Apple and Android mobile app stores were systematically searched for dermatology-related apps that use AI algorithms. Each app's purpose, target audience, evidence-based claims, algorithm details, data availability, clinician input during development, and data usage privacy policies were evaluated. Findings: A total of 909 apps were initially identified. Following the removal of 518 duplicates, 391 apps remained. Subsequent review excluded 350 apps due to nonmedical nature, non-English languages, absence of AI features, or unavailability, ultimately leaving 41 apps for detailed analysis. The findings revealed several concerning aspects of the current landscape of AI apps in dermatology. Notably, none of the apps were approved by the US Food and Drug Administration, and only 2 of the apps included disclaimers for the lack of regulatory approval. Overall, the study found that these apps lack supporting evidence, input from clinicians and/or dermatologists, and transparency in algorithm development, data usage, and user privacy. Conclusions and Relevance: This cross-sectional study determined that although AI dermatology mobile apps hold promise for improving access to care and patient outcomes, in their current state, they may pose harm due to potential risks, lack of consistent validation, and misleading user communication. Addressing challenges in efficacy, safety, and transparency through effective regulation, validation, and standardized evaluation criteria is essential to harness the benefits of these apps while minimizing risks.

Deep learning-aided decision support for diagnosis of skin disease across skin tones.

Although advances in deep learning systems for image-based medical diagnosis demonstrate their potential to augment clinical decision-making, the effectiveness of physician-machine partnerships remains an open question, in part because physicians and algorithms are both susceptible to systematic errors, especially for diagnosis of underrepresented populations. Here we present results from a large-scale digital experiment involving board-certified dermatologists (n = 389) and primary-care physicians (n = 459) from 39 countries to evaluate the accuracy of diagnoses submitted by physicians in a store-and-forward teledermatology simulation. In this experiment, physicians were presented with 364 images spanning 46 skin diseases and asked to submit up to four differential diagnoses. Specialists and generalists achieved diagnostic accuracies of 38% and 19%, respectively, but both specialists and generalists were four percentage points less accurate for the diagnosis of images of dark skin as compared to light skin. Fair deep learning system decision support improved the diagnostic accuracy of both specialists and generalists by more than 33%, but exacerbated the gap in the diagnostic accuracy of generalists across skin tones. These results demonstrate that well-designed physician-machine partnerships can enhance the diagnostic accuracy of physicians, illustrating that success in improving overall diagnostic accuracy does not necessarily address bias.

Large Language Models in Medicine: The Potentials and Pitfalls : A Narrative Review.

Large language models (LLMs) are artificial intelligence models trained on vast text data to generate humanlike outputs. They have been applied to various tasks in health care, ranging from answering medical examination questions to generating clinical reports. With increasing institutional partnerships between companies producing LLMs and health systems, the real-world clinical application of these models is nearing realization. As these models gain traction, health care practitioners must understand what LLMs are, their development, their current and potential applications, and the associated pitfalls in a medical setting. This review, coupled with a tutorial, provides a comprehensive yet accessible overview of these areas with the aim of familiarizing health care professionals with the rapidly changing landscape of LLMs in medicine. Furthermore, the authors highlight active research areas in the field that promise to improve LLMs' usability in health care contexts.

Black Box Warning: Large Language Models and the Future of Infectious Diseases Consultation.

Large language models (LLMs) are artificial intelligence systems trained by deep learning algorithms to process natural language and generate text responses to user prompts. Some approach physician performance on a range of medical challenges, leading some proponents to advocate for their potential use in clinical consultation and prompting some consternation about the future of cognitive specialties. However, LLMs currently have limitations that preclude safe clinical deployment in performing specialist consultations, including frequent confabulations, lack of contextual awareness crucial for nuanced diagnostic and treatment plans, inscrutable and unexplainable training data and methods, and propensity to recapitulate biases. Nonetheless, considering the rapid improvement in this technology, growing calls for clinical integration, and healthcare systems that chronically undervalue cognitive specialties, it is critical that infectious diseases clinicians engage with LLMs to enable informed advocacy for how they should-and shouldn't-be used to augment specialist care.

Session Introduction: Artificial Intelligence in Clinical Medicine: Generative and Interactive Systems at the Human-Machine Interface.

Artificial Intelligence (AI) models are substantially enhancing the capability to analyze complex and multi-dimensional datasets. Generative AI and deep learning models have demonstrated significant advancements in extracting knowledge from unstructured text, imaging as well as structured and tabular data. This recent breakthrough in AI has inspired research in medicine, leading to the development of numerous tools for creating clinical decision support systems, monitoring tools, image interpretation, and triaging capabilities. Nevertheless, comprehensive research is imperative to evaluate the potential impact and implications of AI systems in healthcare. At the 2024 Pacific Symposium on Biocomputing (PSB) session entitled "Artificial Intelligence in Clinical Medicine: Generative and Interactive Systems at the Human-Machine Interface", we spotlight research that develops and applies AI algorithms to solve real-world problems in healthcare.

Auditing the inference processes of medical-image classifiers by leveraging generative AI and the expertise of physicians.

The inferences of most machine-learning models powering medical artificial intelligence are difficult to interpret. Here we report a general framework for model auditing that combines insights from medical experts with a highly expressive form of explainable artificial intelligence. Specifically, we leveraged the expertise of dermatologists for the clinical task of differentiating melanomas from melanoma 'lookalikes' on the basis of dermoscopic and clinical images of the skin, and the power of generative models to render 'counterfactual' images to understand the 'reasoning' processes of five medical-image classifiers. By altering image attributes to produce analogous images that elicit a different prediction by the classifiers, and by asking physicians to identify medically meaningful features in the images, the counterfactual images revealed that the classifiers rely both on features used by human dermatologists, such as lesional pigmentation patterns, and on undesirable features, such as background skin texture and colour balance. The framework can be applied to any specialized medical domain to make the powerful inference processes of machine-learning models medically understandable.

Principles, applications, and future of artificial intelligence in dermatology.

This paper provides an overview of artificial-intelligence (AI), as applied to dermatology. We focus our discussion on methodology, AI applications for various skin diseases, limitations, and future opportunities. We review how the current image-based models are being implemented in dermatology across disease subsets, and highlight the challenges facing widespread adoption. Additionally, we discuss how the future of AI in dermatology might evolve and the emerging paradigm of large language, and multi-modal models to emphasize the importance of developing responsible, fair, and equitable models in dermatology.

Generation of a Melanoma and Nevus Data Set From Unstandardized Clinical Photographs on the Internet.

Importance: Artificial intelligence (AI) training for diagnosing dermatologic images requires large amounts of clean data. Dermatologic images have different compositions, and many are inaccessible due to privacy concerns, which hinder the development of AI. Objective: To build a training data set for discriminative and generative AI from unstandardized internet images of melanoma and nevus. Design, Setting, and Participants: In this diagnostic study, a total of 5619 (CAN5600 data set) and 2006 (CAN2000 data set; a manually revised subset of CAN5600) cropped lesion images of either melanoma or nevus were semiautomatically annotated from approximately 500 000 photographs on the internet using convolutional neural networks (CNNs), region-based CNNs, and large mask inpainting. For unsupervised pretraining, 132 673 possible lesions (LESION130k data set) were also created with diversity by collecting images from 18 482 websites in approximately 80 countries. A total of 5000 synthetic images (GAN5000 data set) were generated using the generative adversarial network (StyleGAN2-ADA; training, CAN2000 data set; pretraining, LESION130k data set). Main Outcomes and Measures: The area under the receiver operating characteristic curve (AUROC) for determining malignant neoplasms was analyzed. In each test, 1 of the 7 preexisting public data sets (total of 2312 images; including Edinburgh, an SNU subset, Asan test, Waterloo, 7-point criteria evaluation, PAD-UFES-20, and MED-NODE) was used as the test data set. Subsequently, a comparative study was conducted between the performance of the EfficientNet Lite0 CNN on the proposed data set and that trained on the remaining 6 preexisting data sets. Results: The EfficientNet Lite0 CNN trained on the annotated or synthetic images achieved higher or equivalent mean (SD) AUROCs to the EfficientNet Lite0 trained using the pathologically confirmed public data sets, including CAN5600 (0.874 [0.042]; P = .02), CAN2000 (0.848 [0.027]; P = .08), and GAN5000 (0.838 [0.040]; P = .31 [Wilcoxon signed rank test]) and the preexisting data sets combined (0.809 [0.063]) by the benefits of increased size of the training data set. Conclusions and Relevance: The synthetic data set in this diagnostic study was created using various AI technologies from internet images. A neural network trained on the created data set (CAN5600) performed better than the same network trained on preexisting data sets combined. Both the annotated (CAN5600 and LESION130k) and synthetic (GAN5000) data sets could be shared for AI training and consensus between physicians.

Large language models propagate race-based medicine.

Large language models (LLMs) are being integrated into healthcare systems; but these models may recapitulate harmful, race-based medicine. The objective of this study is to assess whether four commercially available large language models (LLMs) propagate harmful, inaccurate, race-based content when responding to eight different scenarios that check for race-based medicine or widespread misconceptions around race. Questions were derived from discussions among four physician experts and prior work on race-based medical misconceptions believed by medical trainees. We assessed four large language models with nine different questions that were interrogated five times each with a total of 45 responses per model. All models had examples of perpetuating race-based medicine in their responses. Models were not always consistent in their responses when asked the same question repeatedly. LLMs are being proposed for use in the healthcare setting, with some models already connecting to electronic health record systems. However, this study shows that based on our findings, these LLMs could potentially cause harm by perpetuating debunked, racist ideas.

Skin Tone Analysis for Representation in Educational Materials (STAR-ED) using machine learning.

Images depicting dark skin tones are significantly underrepresented in the educational materials used to teach primary care physicians and dermatologists to recognize skin diseases. This could contribute to disparities in skin disease diagnosis across different racial groups. Previously, domain experts have manually assessed textbooks to estimate the diversity in skin images. Manual assessment does not scale to many educational materials and introduces human errors. To automate this process, we present the Skin Tone Analysis for Representation in EDucational materials (STAR-ED) framework, which assesses skin tone representation in medical education materials using machine learning. Given a document (e.g., a textbook in .pdf), STAR-ED applies content parsing to extract text, images, and table entities in a structured format. Next, it identifies images containing skin, segments the skin-containing portions of those images, and estimates the skin tone using machine learning. STAR-ED was developed using the Fitzpatrick17k dataset. We then externally tested STAR-ED on four commonly used medical textbooks. Results show strong performance in detecting skin images (0.96 ± 0.02 AUROC and 0.90 ± 0.06 F1 score) and classifying skin tones (0.87 ± 0.01 AUROC and 0.91 ± 0.00 F1 score). STAR-ED quantifies the imbalanced representation of skin tones in four medical textbooks: brown and black skin tones (Fitzpatrick V-VI) images constitute only 10.5% of all skin images. We envision this technology as a tool for medical educators, publishers, and practitioners to assess skin tone diversity in their educational materials.