Покоряем новые научные вершины 🚀
Принята статья к публикации и презентации на конференции ICMLC 2026 в Китае 🇨🇳
📍 Nanjing, China
📅 6-9.02.2026
🛜 online
▶️ Design of a Multi-Agent Decision Support System for Handling Patient Complaints
Patient complaints are a valuable source of feedback for healthcare quality improvement, however, handling them manually is labor-intensive and time-consuming. This paper presents a patient-oriented multi-agent decision support system to automate the processing of patient complaints at Medical Information and Analytical Centers (MIAC) in Russia. Seven specialized AI-driven agents were designed using over 120,000 complaints from 2020–2024 submitted via the Gosuslugi e-government portal: (1) text preprocessing, (2) complaint classification, (3) retrieval of knowledge via a vector database and retrieval-augmented generation (RAG), (4) electronic health record (EHR) service validation, (5) information gathering from responsible organizations through chatbots, (6) generative response drafting with explainable AI (xAI) principles, and (7) statistical analysis. The proposed system aims to reduce MIAC staff workload, standardize response quality, and accelerate complaint resolution. Preliminary experiment was conducted evaluating the effectiveness of five contemporary large language models in the task of thematic classification of citizen appeals. The multi-agent approach can substantially assist human experts in complaint management, leading to faster resolutions and actionable insights for healthcare quality improvement, while retaining necessary human oversight for complex cases. Future work will focus on comprehensive real-world evaluation, integration with medical information systems, and continuous learning to further improve performance and transparency.
