Date：22 July (Thursday)
Time: 12:20 – 13:50 (GMT+8)
Chief, Department of Critical Care Medicine
Taichung Veterans General Hospital
The project “Zoe: AI+HI, AI Diagnosis and Caring System and Cross-Hospital Implementation” was proposed by the team composed of Taichung Veterans General Hospital (TCVGH), Tunghai University, and Advantech Co. Ltd. in May, 2021. The team obtained the grant from Ministry of Science and Technology, Republic of China and conducted the 3 different sub-projects. The first is to set the prioritized clinical problems (the prediction model of Acute Respiratory Distress Syndrome, Acute kidney Injury, Bacteremia and Pain) by the physicians of critical care medicine, renal disease and infection disease. The 2nd sub-project is to establish the TCVGH data bank of critical care from 2015~2019 by the engineers of Computer & Communication Centers. The engineers from AI center of Tung Hai University develop the AI prediction model with machine learning. Advantech Co. Ltd provides the platform of Wise-PaaS for the deployment of the AI models and establish the dashboard for real-time prediction. All the members of the 3 sub-teams co-work closely and forms the mutually supportive triangle. We have finished the first-year goals of the project. The clinicians can view the results of real-time prediction and response to the AI prediction. TCVGH AI+HI Smart Critical Care System, enabled by AI and medical expertise, will help those seriously ill patients.
The project “AI+HI Smart Critical System” was conducted by the Taichung Veterans General Hospital (TCVGH), Tunghai University, and Advantech Co. Ltd. from May 2021. We set the four prioritized clinical problems to develop the prediction models of acute respiratory distress syndrome (ARDS), Acute Kidney Injury (AKI), Bacteremia/mortality, and Pain. The Computer & Communication Centers engineers successfully established TCVGH data bank of critical care in 2015~2020. There are 339 features and 66 million records in the data bank. The engineers from the AI center of Tunghai University develop the AI prediction model with machine learning, including ARDS, AKI, and Bacteremia. Advantech Co. Ltd develops Deep SOFA model to predict mortality after 72 hours and set up the platform of Wise-PaaS for the deployment of the AI models. We also establish the interactive dashboard of those predicting models, which shows the real-time prediction result. We have finished the first-year goals of the project. The clinicians can view the results of real-time prediction and response to the AI prediction. Next, we will construct the Federal Learning System of TCVGH AI+HI Smart Critical Care System with other medical centers. We also monitor the impact of deploying those AI models on clinical outcomes. We hope TCVGH AI+HI Smart Critical Care System is enabled by AI and medical expertise and helps those seriously ill patients.