Harnessing AI for Dermatology Residents

Dermatology residents face the dual challenge of delivering high-quality patient care while continually advancing their medical knowledge and research skills. Large language models (LLMs) such as ChatGPT, Claude, Gemini, and Perplexity, along with tools like Google’s NotebookLM, offer innovative solutions to support these demands. Additionally, clinical decision support systems like Glass Health, OpenEvidence, ClinicalKey, and PathwayMD, as well as artificial intelligence (AI)-powered medical scribes such as Abridge or Nuance’s DAX Copilot, are transforming information management and patient care. However, it’s crucial to approach these tools with a critical eye, ensuring the accuracy of the information they provide.

Educational Aides

LLMs can significantly enhance the educational experience for dermatology residents. ChatGPT and Claude, for example, can generate comprehensive explanations of complex dermatological conditions, assist in creating study materials, and simulate patient interactions for training purposes. By inputting specific queries or case scenarios, residents can receive detailed, tailored information that aids in understanding and retention. Google’s NotebookLM further augments this by allowing residents to organize and query their documents and notes efficiently. This tool enables the consolidation of lecture notes, research articles, and personal annotations into a searchable database, facilitating quick retrieval of information and seamless integration into daily learning and practice.

Research Tools

In the realm of research, LLMs like Gemini and Perplexity can assist in literature reviews by summarizing existing studies, identifying research gaps, and even suggesting potential methodologies. These models can process vast amounts of data, providing residents with synthesized insights that inform and streamline the research process. NotebookLM complements this by organizing research materials, tracking sources, and managing citations, thereby enhancing the efficiency and organization of research activities.

In the Clinic and At the Office

Clinical decision support tools such as Glass Health, OpenEvidence, ClinicalKey, and PathwayMD offer evidence-based recommendations and diagnostic assistance, aiding residents in making informed decisions and ensuring that patient care aligns with the latest medical guidelines and research. AI-powered medical scribes, such as Abridge, assist in documentation by transcribing and summarizing patient encounters, reducing administrative burdens and allowing residents to focus more on patient interaction and care.

These AI-powered medical scribe tools offer dermatology residents the potential to streamline documentation by transcribing patient interactions and summarizing key points for electronic health records. This can reduce administrative burdens and allow for more focused patient care. However, it’s essential for residents to consult with their residency program or institution before implementing these technologies, as policies regarding the use of AI tools can vary. Ensuring compliance with institutional guidelines and patient privacy regulations is crucial when integrating AI scribes into clinical practice.

Verify, Verify, Verify

While these AI tools offer substantial benefits, it’s imperative to verify the information they provide. LLMs can sometimes produce “hallucinations,” generating incorrect or fabricated information. To mitigate this, techniques like retrieval-augmented generation (RAG) are employed. RAG enhances the responses of LLMs by incorporating external, up-to-date information. Instead of relying solely on the data within the model’s training set, RAG allows the model to access and reference external sources, such as medical databases or recent publications, before generating a response. This approach grounds the AI’s outputs in current and specific information, thereby reducing the likelihood of inaccuracies.

To utilize RAG with platforms like ChatGPT, the model is integrated with an information retrieval system that fetches relevant data from external sources. When a user inputs a query, the system retrieves pertinent information, which the LLM then incorporates into its response. This process ensures that the AI’s answers are informed by the most recent and relevant data available.

As the medical field rapidly evolves, proficiency with technologies like LLMs, clinical decision support systems, and AI-powered documentation aids will soon be a critical component of modern medical education and patient care. By embracing these innovations now, residents can enhance their learning, streamline research, and improve clinical outcomes, positioning themselves at the forefront of dermatological practice.

Ramie Aly Fathy, MD

• PGY-3, Department of Dermatology, Johns Hopkins University
  Baltimore, MD