Department of Internal Medicine, Division of Cardiovascular Medicine-
Vice Superintendent Kuan-Cheng Chang
AI Center for Medical Diagnosis Director Kai-Cheng Hsu
Acute Myocardial Infarction and Prehospital 12-Lead ECG with AIoT for Arrhythmia Remote Diagnosis
As medical development advances and the average age eventually ages, cardiovascular related diseases have become the important issues in jeopardizing human health. Among which, Myocardial Infarction (MI) is one of the diseases with highest risk since maybe patients had out-of-hospital cardiac arrest (OHCA). Hence, early discovery and diagnosis of patients with myocardial infarction is a very valuable issue in clinical care.
China Medical University Hospital-Vice Superintendent Kuan-Cheng Chang expressed that ECG is a considerably important reference for diagnosing such disease. When the ST segment of ECG revealing specificity elevates (STEMI), it means that the corresponding coronary artery has been completely clogged or nearly completely clogged. When the cardiomyocyte of transmural ventricular wall is found hypoxia and necrosis, failure to recover the blood flow quickly will result in the permanent damage to the heart and higher risks of life-threatening arrhythmia and acute complications such as cardiogenic shock, cardiac perforation and cardiac tamponade. To solve such clinical problem, our research team has established the “acute ST-elevation myocardial infarction STEMI” interpretation model with more precision than the existing ECG expert system in Taiwan, using Artificial Intelligence (AI) and machines. The model combines a 12-lead interpretation for arrhythmia and an automatic high risk scoring system for emergency triage, which becomes a comprehensive acute myocardial infarction ECG diagnosis platform. The initial results have effectively shortened the D2B (door to balloon) upon arrival to the hospital, namely the time to recover myocardial perfusion.
Nonetheless, the improvement of D2B time only pays attention to the medical treatment after the patient arrives to the hospital but in fact the time myocardial ischemia begins is when the patient starts to show symptoms. Therefore, how to shorten the time for patients with myocardial infarction from having symptoms to recovering the blood flow, in other words, S2B (symptom to balloon time) time, has drawn increasing attention in recent years. Other than providing proper health education that will remind patients with awareness for their symptoms, are there other means for patients to determine myocardial infarction early during the symptoms, in other worlds, before arriving to the hospital. On one hand, patients will be warned of early visitation to the hospital and on the hand, such information will be notified to the frontline medical staff and cardiac catheter team upon arrival to the hospital, thereby reducing the unnecessary time consumption, shortening the time for cardiomyocyte hypoxia and retaining heart function.
Moreover, many rural and remote areas in Taiwan still face with severe shortage in medical staff, and patients may not be instantly and correctly interpreted due to lack of professional physicians, even if they take the ECG test. Therefore, how to transmit ECG remotely to allow artificial intelligence automated interpretation will be the most crucial factor in shortening S2B time and improving medical care in remote areas.
The hospital integrates artificial ECG algorithm and small personal ECG instrument, which is incorporated into a simple ECG patch that can be operated by the general public, in attempt to establish the fast ECG remote automatic diagnosis system before patients arrive to the hospital and to shorten S2B (Symptom to Balloon) time. The significance lies on the follows completed outside the hospital: 1. On-site rescue at home or before arriving to the hospital, 2. on the ambulance before arriving to the hospital, and 3. in remote and rural areas lacking regional remote ECG diagnosis. The solution will shorten the time from patients having symptoms to the diagnosis of myocardial infarction or arrhythmia. The combination of personal simple ECG device and AI algorithm provides fast and correct information, so that professional medical staff or ER and cardiology specialist can provide the patients or frontline rescue staff with proper medical suggestions accordingly. Moreover, the hospital can prepare related medical treatment to reduce the damage to the heart and relevant complications as a result of time delay.
China Medical University Hospital AI Center for Medical Diagnosis Director Cheng-Kao Hsu pointed out that the establishment of this AI model integrates the department of cardiology, emergency room, AI Center for Medical Diagnosis, and the Information Office. About 1000 ECG data of patients with myocardial infarction and 2000 ECG data of normal patients at Chinese Medical University Hospital from 2008 to 2018 have been collected to train the AI model. After the completion of model training, the model was launched in emergency room in June, 2020, which has undergone 20,000 ECG on clinical tests with an accuracy rate reaching 99.7%. Currently this AI Model is shifted to extension in fields outside the hospital, which has completed the software and hardware integration and can apply the AI assisted ECG evaluation on ambulances and remote areas.
※ China Medical University Hospital
Introduction to AI Center for Medical Diagnosis
The Center is committed in the AI application of clinical data, which transforms big data into artificial intelligent model with substantial functions through machinery/in-depth learning technology. The Center offers secondary expert opinion on diagnosis when the clinical specialist physicians are conducting medical diagnosis, in addition to incorporating professional medical decisions, thereby reducing the burden on physician from analyzing big medical data. CMUH AI Center for Medical Diagnosis Director Cheng-Kai Hsu stated the four topics of research: medical imaging, bioelectric signals Electronic Medical Record (HER), and genome data undergo artificial intelligent development application. The combination or AI and big data establishes a complete smart solution for diagnosis, treatment and prognosis.