When using a smartwatch to identify cardiac arrest

When using a smartwatch to identify cardiac arrest, 1 in 4 cardiac arrests may be detected, but the potential is missing out on lives saved. SUBHEAD

Detection of Sudden Loss of Pulse with High Accuracy Machine Learning Shows Promise As technology revolutionizes the healthcare industry, Google Research has recently found that wearable technology, for the first time, can track out- of -hospital cardiac arrests (OHCA).

As reported recently in a study in Nature, the significant potential of a smartwatch with a machine learning algorithm capable of independently recognizing sudden loss of pulse and contacting emergency services was highlighted.

The impressive specificity and moderate sensitivity of the technology could lead to improved survival rates, especially in cases of unwitnessed cardiac arrests.

Every year in the United States, nearly 300,000 people die from out-of-hospital cardiac arrest, a critical condition that is often fatal but can also be treated effectively if the right steps are taken immediately after its onset.

In fact, studies have found that the majority of these incidents about 50-75% are unwitnessed, with the victim failing to get immediate help or resuscitation.

The researchers sought to determine whether a consumer-grade smartwatch could automatically recognize these types of life threatening occurrences and summon emergency services, including when the subject is unresponsive.

Google Research has recently found that wearable technology, for the first time

Researchers aimed to strike a balance between minimizing false alarms while making sure there was a reliable and timely response. Conducted by a research team bold tagline Automated Loss of Pulse Detection on a Consumer Smartwatch, the study used a combination of photoplethysmography (PPG) and motion data.

Data from several controlled and real-world settings, including clinical and free-living settings, were used to train the machine learning algorithm. This also allowed the research team to evaluate the accuracy of the system to detect cardiac events in a variety of real-world situations.

To measure how false positives might show up in real life

The researchers used various methodologies to assess the smartphone's ability to detect pulselessness. One study involved 100 patients undergoing defibrillator testing in an electrophysiology laboratory (20), in whom the pulse was purposefully interrupted by inducing ventricular fibrillation, yielding valuable information regarding the detection of pulselessness.

A perm-need for another 99 participants to mimic pulselessness via a tourniquet-induced arterial occlusion model.

They also collected free-living data from a larger cohort of 948 users who did not have any pulseless events. This large amount of data helped to guarantee that the algorithm was capable of operating well under a variety of conditions. To measure how false positives might show up in real life, a total of 220 participants wore the smartwatch as part of the study.

A further 135 participants were observed during free living conditions, where they could experience normal movements alongside the controlled environment where their pulse was intentionally stopped using arterial occlusion.

By combining both methods, it was possible to establish a comprehensive assessment of the specificity and sensitivity of the system. The study also tested the smartwatch's detection accuracy in high-motion scenarios.

The complex nature of real-world cardiac arrests.

To do so, 21 stunt persons trained professionally set up out-of-hospital cardiac arrest collapses so the researchers could see how well the algorithm works under high-pressure conditions.

Crucially, the smartwatch achieved a 99.99% specificity rate, which means that false alarms were rare. On average, for every 21.67 user-years, the algorithm made a false emergency call.

The sensitivity of the algorithm,its ability to accurately identify when the heart's pumping had stopped,was somewhat more variable. Sensitivity for motionless pulseless events was 72% and was lower for simulated collapse events at 53%.

While they described the high specificity and relatively good sensitivity, the authors noted that the algorithm's training on controlled pulseless events may not fully represent the complex nature of real-world cardiac arrests.

Although promising, the smartwatch was able to detect cardiac events within 57 seconds and also trigger an emergency call following a 20-second prompt asking the user if it was indeed a real event, but refining the algorithm to recognise a wider scope of real-life scenarios is needed.

This technology has considerable advantages

So, on-going data collection will further enhance the accuracy and reliability of the algorithm, especially in non-static or dynamic environments. This technology has considerable advantages, one of the biggest being its speed of action in unwitnessed cardiac arrest or in remote areas where the emergency response time may be lengthy.

If this technology is perfected, it could prove invaluable for improving survival rates by rapidly alerting emergency services and guiding bystanders on what to do to help the victim until professional help arrives.

But, the next stage of research deals with the false-positive rate. Like any safety-critical automated system, this algorithm must not generate an emergency call by mistake. Repeated false alarms could result in inefficiencies in emergency response systems and possibly decreased public trust in the technology.

Thank you so much for reading my work.

Any feedback or support that you have to offer is accepted and appreciated