Digital health was a topic of discussion this week at the Heart Rhythm Society Scientific Sessions in Boston as researchers and clinicians from around the world presented their most recent study results. The focus of these studies included wearables, AI, and web-based platforms. 

One new study presented at the event found that popular consumer wearables can accurately measure induced paroxysmal supraventricular tachycardia (PSVT), which means a rapid heart rate above 100 beats per minute. Participants in the study wore either an Apple Watch, Samsung Galaxy Gear, or Fitbit Charge 2. 

Researchers found that the the Apple Watch and Samsung Galaxy Gear were 100 accurate at detecting baseline heart rate within five beats per minute, while the Fitbit Charge 2 was 94 percent accurate. When a participant’s heart rate was in PSVT, the Apple Watch and Samsung Galaxy Gear could accurately detect baseline heart rate within 10 beats per minute 100 percent of the time, and Fitibit could detect it 88 percent of the time. 

The study looked at 51 consecutive patients with a history of PSVT or paroxysmal palpitations who were scheduled to undergo an electrophysiology study. Participants were randomly assigned to wear two of the three watches on each wrist to measure simultaneously. 

“Millions of people around the world are wearing smart watches or other devices that have the ability to track heart rate, and our study is the first of its kind to assess how effective and accurate these tools are for patients,” Dr. Jongmin Hwang, lead author of the study and physician at Seoul Asan Medical Center, Seoul, South Korea, said in a statement. “As clinicians, we see these devices as a tool to help patients learn more about their heart health and become more proactive about self-care. With technology delivering information literally to their fingertips, we hope patients will be better informed and more inclined to speak with their doctor about their health.”

In another study, the Mayo Clinic teamed up with AliveCor to focus on how artificial intelligence and deep neural networks can help identify patients with congenital Long QT syndrome despite normal QTs on their electrocardiogram. 

The Mayo Clinc study found that implementing AI technology generated an area under the curve of 0.83, with a specificity of 81 percent, sensitivity of 73 percent, and overall accuracy of 79 percent at detecting the condition.  

Researchers applied AI to lead one of a 12-lead ECGs. In a statement AliveCore said that this could mean its KardiaMobile and KardiaBand devices could also someday be helpful in detecting patients with LQTS. 

In July 2017 AliveCor and the Mayo Clinic had announced that they were going to work together to use AI to detect LQTS. 

"There can be no better illustration of the importance of our AI to medical science than using it to detect that which is otherwise invisible," Vic Gundotra, CEO of AliveCor, said in a statement.
 
A third study was also released at the conference that looked at online ECG monitoring versus the traditional offline Holter and patch method of detecting paroxysmal arrhythmia. Results from the study showed that the online web platform outperformed offline methods. 

The online platform, called Pocket ECG, had a diagnostic yield six times higher than Holter monitoring in the first 24 hours, and 3.5 times higher than Holter monitoring in the first 48 hours when examining patients with an atrial fibrillation burden of under one percent or lower. Its diagnostic yield was also higher than the offline patch.

"Paroxysmal atrial fibrillation is an arrhythmia that is often difficult to detect because of its episodic nature but presents real health risks to patients," Marek Dzuibinski, lead author and inventor of the PocketECG mobile arrhythmia monitoring solution, said in a statement. "This large-scale study demonstrated that the ability to shorten or extend monitoring duration based on the ongoing results transmitted by online ECG monitoring can improve diagnostic yield over fixed offline methods."