The Department of Critical Care at Taichung Veterans General Hospital (TVGH) has combined medicine and information engineering to develop a number of disease prediction modules using artificial intelligence to assist medical practice and implement them in critical care to improve the quality and effectiveness of patient care.
The critical care team at TVGH has published 12 papers and received 7 patents in Taiwan. The team has also participated in the 25th SNQ Award and won the Silver in the Smart Healthcare category.
In addition to displaying the risk level on the ward dashboard, the team also presented the patient's critical care-related data from the time of admission, including respiratory parameters and fluid status. The treatment of acute respiratory distress includes parameters such as ventilation, fluid balance and nutrition. If the dashboard indicates that the patient has achieved all of the above (green light), it means that the patient's clinical status is gradually improving.
Since the establishment of TVGH's critical care database, it has been a major success. Starting with digitisation, through the implementation of 3C technology, the hospital has built automated data cleaning tools from medical equipment connection, data collection to data cleaning. The database is expected to surpass MIMIC-IV by 2024 and become the world's largest critical care database.
TVGH has developed a critical illness management dashboard, which covers the current status of critical care beds throughout the hospital, the use of important equipment (e.g. Ipecac, ventilators, etc.) and the predicted risk of each patient's different illnesses, providing a quick and efficient overview of the condition from the user's perspective. The dashboard has also been developed respectively for acute respiratory distress syndrome and acute kidney injury. It integrates important clinical data to provide real-time model prediction results, and is visualised to facilitate the understanding of the condition and provide appropriate treatment options.
