About

Roxana Daneshjou is an Assistant Professor of Biomedical Data Science and of Dermatology at Stanford University. She is affiliated with the Center for Artificial Intelligence in Medicine & Imaging (AIMI). Her research focuses on the application of artificial intelligence in medicine and imaging, particularly in dermatology. She contributes to advancing AI-driven healthcare solutions and is involved in academic activities related to biomedical data science and dermatology at Stanford.

Research topics

• Artificial Intelligence
• Medicine
• Computer Science
• Political Science
• Machine Learning
• World Wide Web
• Computational biology
• Risk analysis (engineering)
• Genetics
• Pharmacology
• Pathology
• Medical education
• Biology
• Psychiatry
• Psychology
• Algorithm
• Applied psychology

Selected publications

• Artificial intelligence-enhanced skin self-examinations for skin cancer detection- potential benefits and challenges: insights from an international multidisciplinary expert meeting

  European Journal of Dermatology · 2025-08-01

  article

  International audience

  PublisherDOI

• Expert-level validation of AI-generated medical text with scalable language models

  Research Square · 2025-07-08 · 1 citations

  preprintOpen access
  PublisherOA PDFDOI

• Structured Prompts Improve Evaluation of Language Models

  arXiv (Cornell University) · 2025-11-25

  preprintOpen access

  As language models (LMs) are increasingly adopted across domains, high-quality benchmarking frameworks are essential for guiding deployment decisions. In practice, however, frameworks such as Holistic Evaluation of Language Models (HELM) typically evaluate models under a single static prompt configuration, even though model behavior depends strongly on prompt choice. As a result, reported scores can reflect prompt choice as much as model capability. Declarative prompting frameworks such as DSPy offer a scalable way to evaluate models under a set of structured prompting strategies rather than a static prompt configuration. We present a reproducible DSPy+HELM framework for studying how prompt choice impacts reported benchmark outcomes. Using five prompting methods, we evaluate four frontier and two open-source LMs across seven benchmarks against existing HELM baseline scores. By evaluating LMs across a family of prompt configurations, we find that prompt choice can materially impact leaderboard outcomes. In particular, structured prompting improves performance (by 6% on average), alters comparisons (leaderboard rankings shift on 5/7 benchmarks), with most gains coming from introducing chain-of-thought, and little additional benefit from more advanced optimizers. To our knowledge, this is the first study to systematically integrate structured prompting into an established evaluation framework and quantify how prompt choice alone can impact benchmark conclusions. We open-source (i) DSPy+HELM Evaluation (https://github.com/stanford-crfm/helm/pull/3893) and (ii) Prompt Optimization Pipeline (https://github.com/StanfordMIMI/dspy-helm).

  PublisherOA PDFDOI

• Improving dataset transparency in dermatologic Artificial Intelligence using a dataset nutrition label

  npj Digital Medicine · 2025-11-05 · 1 citations

  letterOpen access

  Biased and poorly documented dermatology datasets pose risks to the development of safe and generalizable artificial intelligence (AI) tools. We created a Dataset Nutrition Label (DNL) for multiple dermatology datasets to support transparent and responsible data use. The DNL offers a structured, digestible summary of key attributes, including metadata, limitations, and risks, enabling data users to better assess suitability and proactively address potential sources of bias in datasets.

  PublisherOA PDFDOI

• Using Large Language Models to Audit Model Healthcare Biases

  2025-12-01

  articleOpen accessSenior author

  Large language models (LLMs) can potentially mitigate pain points in healthcare tasks such as decision support, text summarization, and question-answering. However, LLMs exhibit bias related to race, gender identity, sexual orientation, and other demographics, posing a major concern. Although human review helps reduce bias, the sheer data volume renders thorough evaluation impractical and onerous at scale. This motivates the use of LLMs in auditing models for bias. This study uses the Stanford Healthcare red-teaming dataset, which contains prompts, outputs, and expert-level bias labels, to examine how model size and prompting techniques affect bias detection with GPT-3.5-turbo, GPT-4o, llama3.3, and o1-mini. Our results show that the best model for bias detection depends on the chosen metric. Smaller, cost-effective models like o1-mini outperformed GPT-4o in precision and F1 scores, with up to 53.11% higher precision and 10.32% higher F1. This suggests that smaller models may be preferable when precision or F1 is a priority. Additionally, self-critiquing capabilities in larger models do not significantly improve bias detection over smaller models (χ2, p = 0.597). Moreover, the use of prompting techniques, particularly Thread of Thought, significantly enhanced bias detection across all models, (χ2, p < 0.001). Our findings suggest that depending on the metric of concern for the auditor, smaller models can offer a costeffective alternative to larger models.

  PublisherDOI

• The TRIPOD-LLM reporting guideline for studies using large language models: a Korean translation

  The Ewha Medical Journal · 2025-07-31 · 1 citations

  articleOpen access

  대형 언어 모델(large language model, LLM)의 활용이 의료 분야에서 빠르게 확대되면서, 표준화된 보고 지침의 필요성이 커지고 있다. 이 논문에서는 LLM을 활용한 연구를 위한 다변수 예측모델의 투명한 보고(TRIPOD-LLM) 지침을 제시하였다. TRIPOD-LLM은 기존 TRIPOD와 인공지능(artificial intelligence) 확장 지침을 기반으로 하며, 바이오 메디컬 분야에서 LLM이 가지는 고유한 도전 과제들을 반영하고 있다. 이 지침은 제목부터 논의까지 주요 내용을 포괄하는 19개 주요 항목과 50개 세부 항목으로 구성되어 있다. 다양한 LLM 연구설계와 작업에 적용할 수 있도록 모듈형 형식을 도입하였고, 모든 연구에 공통적으로 적용할 수 있는 14개 주요 항목과 32개 세부 항목을 포함한다. 이 지침은 신속한 델파이(Delphi) 과정과 전문가 합의를 거쳐 개발하였으며, 투명성과 인간 감독, 과업 특이적 성과(task-specific performance) 보고의 중요성을 강조한다. 또한 지침의 손쉬운 작성과 제출용 PDF 생성을 지원하는 인터랙티브 웹사이트(https://tripod-llm.vercel.app/)를 소개한다. TRIPOD-LLM은 ‘생명력 있는 문서’로서, 연구현장의 변화에 맞추어 지속적으로 개정될 예정이다. 이 지침을 통해 LLM 연구의 보고 수준을 높이고, 재현성과 임상 적용 가능성을 강화하는 데 기여하려고 한다.

  PublisherDOI

• Automated Detection of Benign and Malignant Skin Lesions from Reflectance Confocal Microscopy Images Using Deep Learning

  JID Innovations · 2025-08-08 · 2 citations

  articleOpen access

  Reflectance confocal microscopy offers a noninvasive approach for diagnosing skin lesions at the point of care, but it remains underutilized owing to the specialized skill required for interpretation. Artificial intelligence provides an opportunity to automate this process. We developed deep learning models to automate the analysis of reflectance confocal microscopy block images. Reflectance confocal microscopy images acquired from 3rd and 4th generation VivaScope 1500 devices were preprocessed and split for training and testing. Two models were developed: a modified convolutional neural network ResNet-18, for skin layer detection, and a ResNet-34 integrated with a gated recurrent unit for lesion classification. The models were pretrained on 3rd generation images and fine tuned on 4th generation data, utilizing 5-fold cross-validation. Our cohort included 845 patients, 1147 lesions, and 4391 VivaBlock images. The layer detection model identified the dermis, epidermis, and dermoepidermal junction, achieving an area under the curve of 0.70, 0.71, and 0.57, respectively. The lesion classification model distinguished malignant from benign lesions with an area under the curve of 0.80 and specificity of 0.91. Our convolutional neural network gated recurrent unit approach effectively distinguished benign from malignant lesions, showing impressive diagnostic accuracy mimicking expert dermatological assessments. This highlights artificial intelligence's potential in improving reflectance confocal microscopy image interpretation, reducing unnecessary biopsies, and paves the way for future research.

  PublisherDOI

• Policy brief: AI-first Medicaid: how CMS can build a smarter safety net with Precision Benefits

  npj Digital Medicine · 2025-11-28

  articleOpen accessSenior author

  Medicaid serves over 70 million Americans, yet barriers to consistent, high-quality care endure due to workforce shortages, fragmented service delivery, and administrative burden. Artificial intelligence (AI) offers not just operational efficiency but the potential to transform the Medicaid care experience. AI-powered digital assistants can deliver 24/7 multilingual voice or text support, expanding access to personalized, emotionally-intelligent assistance. Under existing workforce supervision, these agents can bridge critical gaps in behavioral health and community coordination through tools like therapy chatbots that reduce loneliness and improve engagement. As “embedded staff” in provider offices and community organizations, digital assistants can create a unified infrastructure for whole-person care. We introduce the concept of Precision Benefits: delivering the right support to the right person at the right time to prevent avoidable health and social deterioration. This aligns with administrative and eligibility reforms in H.R.1, which require states to improve efficiency and verification while fostering innovation and preserving state authority over AI regulation. Realizing this vision demands responsible AI development – addressing safety, bias, privacy, and trust – and modernization of infrastructure and payment models. Yet the opportunity is clear: AI can power a smarter and more equitable Medicaid system, one that puts everyone on an upward life trajectory.

  PublisherOA PDFDOI

• 64724 Skincare Influence: Exploring Dermatology Trends on Social Media

  Journal of the American Academy of Dermatology · 2025-09-01

  articleSenior author
  PublisherDOI

• Session Introduction: AI and Machine Learning in Clinical Medicine Bridging or Separating Model Intelligence and Human Expertise

  2025-12-01

  articleOpen access

  Artificial Intelligence (AI) technologies continue to expand their role in clinical medicine, with large language models (LLMs) and multimodal systems now applied to communication, imaging, and predictive analytics. Advances in generative and retrieval-augmented methods have improved the accuracy and contextual grounding of clinical summaries, patient messaging, and decision support. At the same time, new benchmarks in imaging, vision, and spontaneous speech have underscored both progress and the persistence of unsolved challenges. Predictive modeling efforts highlight causality, longitudinal trajectories, and informative clinical events, while methodological contributions emphasize uncertainty management, abstention, and interpretable causal structures. Finally, frameworks for evaluation and governance address the crucial gap between laboratory performance and real-world deployment.

  PublisherDOI

Frequent coauthors

• Justin Ko

  101 shared

• Jesutofunmi A. Omiye

  92 shared

• Haiwen Gui

  Stanford University

  91 shared

• Vijaytha Muralidharan

  University of Alberta

  88 shared

• Michael Vollmer

  Hospital Israelita Albert Einstein

  81 shared

• Christopher Hyde

  University of Exeter

  81 shared

• Ping Diao

  Sichuan University

  81 shared

• Márcio R. C. Reis

  Hospital Israelita Albert Einstein

  81 shared

Education

• Ph.D., Biomedical Informatics

  Stanford University

  2016

• B.S., Biology

  University of California, San Diego

  2010