We built an AI model to predict AKI risks in the next 24 hours using TCVGH＇s ICU database. The model was validatedoptimized through federated learning among five medical centers. We developed an interactive dashboardintegrated it into the hospital＇s medical system for real-time inference. Finally, we organized those technologiescreated the "AI Acute Kidney Injury Prediction: Real-time Inference Interactive Intensive Care System."

We integrated five significant technologies to establish an AI prediction model for AKI risk. Through federated learning, the model was jointly optimized with four medical centers. An interactive dashboard is integrated with the hospital information system, enabling real-time data transmissioninteractive feedback to display real-time prediction results. We complete the“Artificial Intelligence Acute Kidney Injury Prediction: Real-Time Inference Interactive Critical Care System.”

This technology is applied to address the challenges of intensive careacute kidney injury (AKI) care needs within a medical information system that meets the requirements. Through the AI inference engine, high-risk patients are marked in real-time, optimizing workflowsassisting physicians to intervene before AKI, thereby enhancing the quality of care. The system can upgrade the intensive care unit＇s monitoring system functionsbe applied to remote intensive care."