Researchers in the United States have used machine learning to identify abnormal cardiac rhythms in firefighters in a move they say could lead to an early warning system and potentially prevent fatal incidents.

Around 40% of fatalities for on-duty firefighters in the US come from sudden cardiac arrest, according to the joint research team from the National Institute of Standards and Technology (NIST), the University of Rochester and Google, who worked together using the AI tool.

The team hopes its study findings, published in Fire Safety Journal, will eventually lead to a portable heart monitor that firefighters could wear to catch early warning signs of heart trouble and prompt them to seek medical attention before a fatal incident occurs.

As the researchers note: ‘Firefighters work in remarkably strenuous environments, carrying heavy objects, climbing stairs, and enduring extreme temperatures, with a limited ability to cool off. And while they may experience significant discomfort… firefighters often try to push through these situations without realising they may be at risk for sudden cardiac death.’

Researchers at the University of Rochester’s School of Nursing had collected 24 hours of electrocardiogram (ECG) data from each of 112 firefighters, who had electrodes strapped to their chests.

As the researchers explain, the data that was collected covered 16-hour on-duty shifts and eight-hour off-duty shifts during which the firefighters engaged in their daily activities such as answering fire and medical calls, exercising, eating, resting and sleeping.

Working with the University of Rochester dataset, the NIST team used machine learning to build what they call the Heart Health Monitoring (H2M) Model and trained it with 12-second segments of a large portion of the ECG data.

As the researchers explain, individual heartbeats in the ECGs were classified as normal beats or abnormal beats indicative of irregular heart rhythms such as an atrial fibrillation or ventricular tachycardia.

Once the model had been trained and validated on this dataset, the researchers used the AI tool to analyse firefighter ECG data collected by the University of Rochester that it had not previously seen.

According to the research paper, when the model was presented with around 6,000 abnormal ECG samples, it correctly identified them with about 97% accuracy.

As a further check, the model was also trained using ECG datasets from non-firefighters. When these were used, the AI tool demonstrated an error rate of around 40% in identifying cardiac events in the firefighter data, according to the study.

By incorporating the AI tool into portable heart monitors that firefighters wear on duty, the researchers argue the early warning system could warn them of cardiac irregularities in real time.

The study also suggests that other groups could also benefit from the AI tool if it was trained with appropriate datasets.

Comment: The technology being created to help firefighters, may prove invaluable in other workplaces and members of the public.

Source: IOSH Magazine