The solution to the silent killer: China Medical University Hospital developed AI sepsis real-time prediction

Source：China Medical University & Healthcare System

The mortality of septic patients in Taiwan was 287 per 100,000 people, with case fatality 29.2%. Sepsis is also known as the "silent killer" since it is difficult to distinguish and with diverse symptoms and multiple causes. Dean Kai Chen, Hus of Artificial Intelligence (AI) Center for Medical Diagnosis of China Medical University Hospital (CMUH) indicate that we uses deep learning technology to develop the " AI -assisted medical diagnosis system for sepsis" to assist in early detection of sepsis and improving survival rate.

Sepsis is an infectious disease with a high mortality rate. An authoritative medical journal "The Lancet" has published the latest statistics of 196 countries around the world by an American research institution: one in every 5 deaths in the world is due to sepsis. In addition, research from Taipei Municipal Wanfang Hospital showed that in Taiwan, there were 643 new cases of sepsis in 100,000 Taiwanese. Director Der-Yang Cho pointed out that traditionally, when an inpatient develops an infection, the physician will firstly determine the possible cause of infection based on his own experience because of emergency purpose, and then administer the corresponding antibiotics. Therefore, a certain proportion of patients are still suffering from the disease. They cannot be diagnosed at the initial stage, so it is easy to miss the golden treatment time.

Dean Kai-Cheng Hsu said that this system has been applied to the "Assisted Medical Decision-Making Platform For Antibiotic Treatment" developed by the hospital's Director Der-Yang Cho and the integrated cross-departmental medical team and this platform integrates three functions of drug administration advice, prediction of drug resistance and sepsis, and has been introducing into clinics. At present, this system has been fully introduced into the medical system of CMUH and expect to be extended to other hospitals in the future to improve the overall quality of critical care for patients with sepsis.

Dean Kai-Cheng Hsu said that the latest paper submitted by the AI Center for Medical Diagnosis in 2021 in the international conference 《CIBB 2021》 pointed out that the system utilized gender, age, vital signs, and CBC data within 3 days of the inpatients, comparing with the past big-data to infer the current probability of sepsis. If the system warns that the possibility of sepsis is high, medical staff can strengthen medical care and vital signs monitoring in advance; in emergency, the medical team can give antibiotics and infusion treatment in time and provide treatment suggestions based on the degree of damage to each organ. Currently, the accuracy of AI predictions trained by CMUH can reach nearly 90%.

CMUH has cooperated with EverFortune.ai Company and started to deploy the US FDA and Taiwan TFDA licenses actively to promote commercial development. Physician of Infectious Medicine Chia-Hui Chou who is work in CMUH said that this system, combined with wearable vital signs monitoring devices and smart wards, has the potential to develop telemedicine in the future. It can help small or remote hospitals with insufficient human resources improve the inspection efficiency, reduce the mortality rate, and solve the problem of long-term emergency congestion.