Session (CTI)

Separate registration required.

Examine large language models (LLMs) from industry leaders such as OpenAI, Anthropic AI, Google, and Meta, focusing on their application in real-world evidence generation and HEOR. The course covers technical LLMs, including their architecture, processing layers, attention mechanisms, embeddings, context window, hallucinations, risk-based frameworks, and current task-specific live benchmarks used for model assessment.

Participants will learn prompt engineering through hands-on, practical examples, empowering them to utilize commercially available LLMs. These examples include scientific literature retrieval, PICO extraction and processing, extracting and handling numerical data, summarizing tables and figures, automating captions, and generating code.

Upon completing this in-depth course, participants will gain the competencies needed to use LLMs responsibly for practical applications in RWE and HEOR, while remaining mindful of regulatory obligations. To participate in practical exercises, attendees are required to bring a personal laptop and have access to a personal or corporate LLM account with file upload functionality.

PREREQUISITE: General knowledge of chat-based LLMs (GPT, Claude, etc) is important. This is an intermediate course, and students should have prior knowledge of AI and have used chat based LLMs in a professional/work setting.