Symptoms of a heart attack may resemble noncardiac symptoms, making diagnosis difficult. British researchers are turning to machine learning to give doctors a way to diagnose heart attacks quickly and accurately, thereby reducing the time needed for diagnosis and making it more efficient and effective. Treatment may be available to patients.
Currently, the leading method of diagnosing heart attacks is to measure the level of the troponin protein in the blood. Troponin is released when the heart muscle is damaged. Levels typically increase sharply within 3 to 12 hours after a heart attack, peaking about 24 hours later.
Many hospitals around the world have adopted diagnostic pathways that include assessment of troponin levels upon admission for suspected heart attack. But they have some limitations. This can be difficult in the emergency department setting as blood samples must be taken over a period of time. They simply classify patients as having low, intermediate, or high risk of heart attack, without considering when symptoms began or other important information such as electrocardiogram (ECG) findings. And the effects of gender, age, and comorbidities were not considered.
British researchers have now developed a fast and accurate AI-based machine learning algorithm. The algorithm, named “Collaboration for the Diagnosis and Assessment of Acute Coronary Syndrome (CoDE-ACS)”, was designed to calculate the probability of heart attack for an individual patient.
The researchers used data from 10,286 patients with possible heart attacks in six countries around the world. A machine-learning algorithm was “taught” using the patient’s gender, age, electrocardiogram findings, and medical history, in addition to troponin levels, to determine the probability of having a heart attack.
The researchers found that CoDE-ACS could rule out heart attacks in more than twice as many patients with an accuracy of 99.6% compared to existing methods.
The algorithm accurately predicted heart attacks across subgroups, including men, women, the elderly, those with kidney (kidney) impairment, and those with early hospital visits after symptom onset.
The researchers say the CoDE-ACS algorithm could prevent unnecessary hospitalizations in patients who are unlikely to have had a heart attack or who are at low risk of myocardial damage or death after a heart attack. This will make emergency care more efficient and effective, they say, and help identify which patients can safely go home and which need to stay for further testing. .
“Early diagnosis and treatment are life-saving for patients with acute chest pain from a heart attack,” said Nicholas Mills, corresponding author of the study. “Unfortunately, many diseases cause these common symptoms, and diagnosis is not always straightforward. Harnessing data and artificial intelligence to support clinical decision-making is an important part of patient care. and has great potential to improve the efficiency of busy emergency departments.”
CoDE-ACS is currently being piloted in Scotland to see if it can reduce the strain on overcrowded emergency departments.
The study was published in a journal natural medicine.
Source: British Heart Foundation
