Clinical Inbox Managers Will Need More Than LLMs to Work | by Karl Swanson | Medium

Amidst the rising blathering about the potential of large language models (LLMs) in the clinical world, we healthcare workers have been inundated yet again with the promises of “AI.” Some of these are grounded in reality, while others are more ambitious (read: dubious). The cacophony of “AI experts” and “thought leaders” dissecting what’s achievable from what’s merely aspirational should mostly be ignored, especially by those who have never built a commercially successful product. However, there’s one promise that stands out, not only for its immediate feasibility but also for the profound difference it could make in clinicians’ lives: leveraging LLMs to manage — *shudders* — “The Inbox” or “The In-Basket.” For those unfamiliar, the inbox is the digital nexus where physicians receive messages directly from patients via “patient portals.” In the most used commercial EHR, Epic, this is called MyChart. Providers can be inundated with hundreds of messages a day, and answering these can cos

(FYI: There is a TLDR at the end if that suits you!) 16 min read Aug 16, 2023 -- Context Matters!! Amidst the rising blathering about the potential of large language models (LLMs) in the clinical world, we healthcare workers have been inundated yet again with the promises of “AI.” Some of these are grounded in reality, while others are more ambitious (read: dubious). The cacophony of “AI experts” and “thought leaders” dissecting what’s achievable from what’s merely aspirational should mostly be ignored, especially by those who have never built a commercially successful product. However,…

related reading

• What We’ve Learned From A Year of Building with LLMs – Applied LLMsapplied-llms.org
• LLM Powered Autonomous Agents | Lil'Loglilianweng.github.io
• Evaluation and mitigation of the limitations of large language models in clinical decision-making | Nature Medicinenature.com
• 2309.07430.pdfarxiv.org
• Large language models are few-shot clinical information extractorsarxiv.org
• Google’s AMIE Validates The Clinician Cockpit | Counsel Healthcounselhealth.com
• Do we still need clinical language models?proceedings.mlr.press
• Clinical Text Summarization: Adapting Large Language Models Can Outperform Human Experts - PMCncbi.nlm.nih.gov
• What We Learned from a Year of Building with LLMs (Part I) – O’Reillyoreilly.com
• Things we learned about LLMs in 2024simonwillison.net
• 2025: The year in LLMssimonwillison.net
• LLMs Are Not All You Need | Pineconepinecone.io