Alex Pavlushkin - Generating Document Drafts as QuestionnaireResponses Using LLMs | DevDays 2024 скачать в хорошем качестве

Alex Pavlushkin - Generating Document Drafts as QuestionnaireResponses Using LLMs | DevDays 2024

Recent strides in Large Language Models (LLMs), combined with the power of FHIR, have created unprecedented opportunities to reduce the workload in overstretched healthcare systems. Our project leverages this advancement to save time for medical professionals, and we're eager to share the results. The most effective approach we've adopted is utilizing LLMs as a co-pilot for healthcare professionals, automating time-intensive tasks like generating drafts of medical documents. Our developed tool proficiently generates valid and detailed QuestionnaireResponse resources, based on the given Questionnaire and various contexts. These contexts range from medical examination audio recordings to textual descriptions, sets of other FHIR resources, medical terminology value sets or their combinations. We've also integrated OpenAI's function calling tools to validate outputs and fetch additional FHIR resources during runtime.