Disorders of the heart and blood vessels are a leading cause of mortality in most countries around the world. Many of those diseases can be diagnosed using ECG, which is a simple and cost-effective tool to measure the activity of the heart. Since the development of this technology over a century ago, many improvements have been made, and lately, a new trend is emerging: ECG readings augmented with AI algorithms. 

ECG readings are captured during various periods of time, depending on diagnostic needs, while some can take only thirty seconds, some are registered over twenty-four hours using devices called Holter monitors. Manually browsing such large quantities of data was inefficient, and so computer analysis of ECG was introduced. However, this approach’s usability became limited, interpretations given by the software were often inaccurate and required a lot of physician attention to give correct diagnosis. 

Machine learning algorithms are said to provide more human-like interpretations of data than classical ECG software. One study found that AI provided more accurate results than conventionally used computer analysis tools, however, it didn’t level with clinician manual interpretation.  

Another interesting case is the usage of AI in heart care on smaller devices equipped with simple ECG, such as smartwatches. Studies demonstrate that AI-ECG algorithms can detect various arrhythmias, which are a group of heart disorders, using electrode configuration that can be embedded into a smartwatch. And it’s not only theory, Samsung has recently announced that they will use AI technology in their Galaxy Watch for arrhythmia detection in daily life.  

One more interesting approach to AI-ECG is the use of Explainable Artificial Intelligence. A study carried out in Japan, showed that it can allow for high accuracy of detection, and identifying interesting regions of longer recordings.  While maintaining those properties, it also allows physicians to familiarize themselves with this technology, and not only see it as a black box. 

As AI-based algorithms still have a lot of room for growth, they seem like a very prospective tool for the future, serving as a help for physicians, allowing them to reduce their workload and create a more robust workflow, while still maintaining high accuracy of a given diagnosis. 



Ryszard Błażej

Student at AGH University of Science and Technology

Zostaw komentarz

Twój adres e-mail nie zostanie opublikowany. Wymagane pola są oznaczone *