Development and Validation of an Integrated Artificial Intelligence-Powered Clinical Decision Support Application for...

This study describes the development and preliminary validation of an integrated clinical decision support application (Cardiac Critical Care Advisor Pro, CCA Pro v6.0) that combines viscoelastic coagulation analysis, arterial blood gas interpretation, mechanical ventilation management, intraoperative hemodynamic assessment, and extracorporeal membrane oxygenation (ECMO) management with artificial intelligence (AI)-powered image recognition in a single platform. This study used software development and descriptive validation to simulate clinical scenarios in a cardiac surgical intensive care unit at a university teaching hospital. The application was evaluated using 50 simulated clinical scenarios derived from real cardiac surgical cases, with expert review by 3 senior cardiac anesthesiologists. Development of a web-based application incorporated 6 clinical modules with 108 input parameters, over 100 automated calculations, evidence-based treatment algorithms with weight-adjusted dosing, and 7 AI-powered camera zones for automated clinical screen data extraction. The application demonstrated agreement with expert consensus in coagulation management (ROTEM/TEG pathway selection and dosing), acid-base classification, ventilator parameter recommendations, and ECMO troubleshooting. The AI camera system successfully extracted clinical values from photographs of ROTEM displays, blood gas printouts, ventilator screens, patient monitors, and ECMO consoles. All treatment recommendations included weight-adjusted dosing consistent with current guidelines (ARDSNet, Surviving Sepsis Campaign 2021, Extracorporeal Life Support Organization 2021, Berlin criteria for acute respiratory distress syndrome). CCA Pro v6.0 represents a comprehensive, integrated clinical decision support tool for cardiac critical care. By combining multiple clinical domains with AI-powered data capture, the application has the potential to reduce cognitive load, minimize dosing errors, and standardize evidence-based practice in high-acuity settings. Prospective clinical validation is warranted.