In August 2022, Chi Mei Medical Center Hospital and SinLau Hospitall signed a "Medical Artificial Intelligence Collaboration Agreement," officially embarking on a journey of AI collaboration. The partnership prioritizes the introduction of Chi Mei Medical Center’s cloud-based AI prediction services for emergency chest pain and pneumonia patients, making these services available in the emergency departments of SinLau Hospitall and Madou SinLau Hospitall. These AI prediction services can forecast the risk of mortality and myocardial infarction within one month for chest pain patients and predict the risks of respiratory failure, sepsis, shock, and death for pneumonia patients.
Chi Mei Medical Center Hospital's Director, Lin Hong-rong, highlighted that unlike other medical centers focusing on imaging AI, Chi Mei Medical Center’s Smart Medical Center leverages its analytical strengths to develop AI predictive models based on electronic medical records. Collaborating closely with the information department, the research outcomes are integrated into the hospital information system (HIS) and applied in clinical settings. To date, 30 types of AI predictive systems have been deployed across three hospital campuses, covering emergency, surgery, anesthesia, intensive care, reproductive medicine, dialysis centers, thoracic medicine, and pediatrics.
Liu Zhong-feng, Director of the Chi Mei Smart Medical Center, cited an example involving an 82-year-old woman with diabetes and coronary artery disease who had previously undergone stent placement. She had stopped her heart medications for three years due to mobility issues. She presented to SinLau Hospitall’s emergency department with sudden chest pain and difficulty breathing. Her blood pressure was low at 102/72 mmHg (normal 120/80 mmHg), her oxygen saturation was 91%, her breathing was rapid (22 breaths/min), her pulse was 84 beats/min, and her ECG showed signs of old myocardial infarction, although her cardiac enzymes were normal.
Given the complexity and severity of her condition, it was unclear if she had acute myocardial infarction. However, using Chi Mei Medical Center Hospital’s chest pain cloud AI prediction tool, it was found that her risk of myocardial infarction within one month was 43.3%, with a mortality rate of 94.64%. Subsequent cardiac catheterization revealed severe blockages in the left main and anterior descending arteries, with left ventricular ejection fraction reduced to 30%. The SinLau Hospitall medical team immediately performed emergency catheterization, stabilizing her condition with vasopressors and addressing atrial fibrillation. Fortunately, the timely intervention restored blood flow, and she was discharged six days later.
The development of medical AI requires vast amounts of medical data and substantial investment in computing infrastructure, posing significant challenges for individual hospitals. Recognizing this, Chi Mei Medical Center Hospital and SinLau Hospitall initiated their AI collaboration in August 2022, introducing Chi Mei Medical Center's AI predictive services for chest pain and pneumonia patients to SinLau Hospitall and Madou SinLau Hospitall’s emergency departments.
Director Liu Zhong-feng noted that both hospitals are certified under ISO27001 ISMS (Information Security Management System). To ensure data network security, cloud computing speed, and information privacy, Chi Mei Medical Center Hospital employs SSL (Secure Sockets Layer) asymmetric encryption and JWT (JSON Web Token) technologies for secure identity verification during data transmission. Given operational differences between the hospitals, AI models were adjusted and optimized for SinLau Hospitall, with each model accompanied by feature importance diagrams (SHAP diagrams) to help users understand key risk factors. After extensive efforts, the chest pain prediction function successfully passed connection tests in October 2023, and the pneumonia prediction function followed in February 2024. This streamlined, secure process sets a viable path for AI sharing in healthcare.
Liu Chi-ju, Director of Tainan SinLau Hospitall, stated that since the AI predictions were implemented, they have provided substantial support to the emergency department, especially aiding less experienced physicians. AI helps them quickly anticipate potential changes in patient conditions and respond appropriately, enhancing response efficiency and resource utilization. He expressed gratitude for Chi Mei Medical Center Hospital’s generous sharing of their AI system. Beyond immediate benefits, the collaboration has allowed SinLau Hospitall staff to experience the advantages of medical AI firsthand, enhancing their proficiency in smart healthcare.
Chi Mei Medical Center Hospital’s Director Lin Hong-rong emphasized that the collaboration with SinLau Hospitall to establish a secure AI cloud sharing mechanism marks a significant milestone in Taiwan’s medical AI development. The success of this model has the potential to reduce the urban-rural healthcare digital divide, advancing medical equity. With continuous improvements in AI model applicability and quality, and the diversification of services, Taiwan is poised to play a crucial role in global smart healthcare development.
Resource (mandarin):
