LLM-Based Framework for Administrative Task Automation in Healthcare

Abstract

Artificial Intelligence (AI) has been transformative in the healthcare sector, leading to enhanced precision in medical diagnosis, more effective treatment options, and a significant improvement in patient safety. However, computer-based administrative tasks, such as retrieval of medical and health records, patient registration, medical billing, filing and documentation, and appointment scheduling, still impose a heavy burden on healthcare professionals, causing a reduced quality of care and efficiency. In light of these challenges, this paper proposes a large language model (LLM)-based multi-agent framework designed to automate some of the administrative work in clinical settings. In our proposed solution, these LLM agents coordinate to parse instructions, breakdown tasks, and execute a sequence of actions in a workflow. They are equipped to not only execute documentation process at the database level but also operate directly on web-based electronic medical record (EMR) platforms. Moreover, the framework integrates data sources through a retrieval-augmented generation (RAG) system to allow streamlined interaction with patient information and medical records, mediated through an agent interface. The framework is designed with security in mind to defend against malicious prompts. We demonstrate the practicality of our solution by testing on various complex tasks that require the use of multiple tools and an EMR website. The result show the framework's effectiveness in handling diverse healthcare administrative tasks.

Ähnliche Arbeiten