AI-based single-lead ECG monitor improves AF recurrence detection, anticoagulation adherence

An artificial intelligence (AI)-based analysis algorithm helps increase the accuracy of a portable single-lead electrocardiogram (ECG) monitor for detecting atrial fibrillation (AF) recurrence, with an added advantage of improving adherence to oral anticoagulation, among patients who underwent catheter ablation, according to a study.

Named BigThumb, the portable heart monitor is a small smartphone-coupled handheld tool that permits easy and rapid collection of a one-lead ECG. Users only have to place their two thumbs on the device. The resulting data will be sent to doctors automatically and stored on the device for review at any time.

The device was tested in a cohort of 218 patients (average age 63 years, 70.6 percent male) with symptomatic AF refractory to at least one class I or class III antiarrhythmic drug and referred for a first catheter ablation procedure. They were randomly assigned to either the BigThumb group (BT Group) or traditional follow-up group (TF Group). There was no statistical difference detected in the demographic data between patients in the two groups.

Patients in BT Group had to take at least three ECGs with the device daily, or more if symptomatic. They were instructed to mark symptoms during the recording. AF was detected by both an automated AF detection algorithm and an AI algorithm, with two cardiologists confirming the diagnosis separately. On the other hand, patients in the TF group used Holter monitors at 3, 6, and 12 months after ablation and ECGs if there are symptoms.

Three months after ablation, the patients were recommended oral anticoagulation if the CHA2-DS2-VASc score was >1. Patients who regularly took anticoagulants during this period were considered adherent.

The BT Group recorded 26,133 ECGs in total, among which 3,299 (12.6 percent) were manually diagnosed as AF by cardiologists. The sensitivity and specificity of the AI algorithm were 94.4 percent and 98.5 percent respectively, which were significantly higher than the values obtained with automated AF detection algorithm (90.7 percent and 96.2 percent, respectively). [Int Heart J 2021;62:786-791]

Use of BigThumb led to more frequent detection of recurrence in paroxysmal AF after ablation (p=0.0099), but not in persistent AF (p=0.7910). Furthermore, significantly more patients in the BT than the TF group showed adherence to oral anticoagulation (p=0.0052).

“In this present study, we found that 785/3,299 (23.8 percent) of the AF detected after ablation were asymptomatic, together with unawareness of AF-related outcomes and consequent comorbidities, which may explain the poor compliance with anticoagulation after catheter ablation even if patients receive recommendations of anticoagulation from doctors,” the investigators pointed out. “The BigThumb may help to record the episode of asymptomatic AF according to random monitoring, which may lead to subsequent adherence to anticoagulation.”

Ready for clinical practice?

Other means to monitor recurrence such as insertable cardiac monitors and cardiac implantable electronic devices can also detect a high rate of AF typically missed during routine clinical care in patients after ablation. This is important, according to the investigators, as early detection of AF has been identified to be crucial in order to define or change proper medical treatment. However, implantation of the devices is invasive and expensive, which prevent them from being used widely. [JACC Clin Electrophysiol 2017;3:1557-1564; Europace 2017;19:1101-1108; J Cardiovasc Electrophysiol 2016;27:1403-1410; PLoS ONE 2019;14:e0216530]

“The BigThumb, [in comparison], is an effective and affordable public ECG monitoring device, which [can] increase the success of incorporating the consumer-generated biometrics into clinical practice,” they added.

However, the device has its limitations, the investigators acknowledged.  “ECG tracings in this clinical trial provide an insight into the real-world limitations of the technology. Because of random and haphazard collection of ECGs, the accuracy of AF detection with the BigThumb highly depends on the compliance of participants.”

The investigators also noted that the device was used more frequently in the first 3 months of follow-up vs afterwards and at daytime vs night-time. So, patients might need to be reminded to persistently used the device.

Still, using BigThumb with AI algorithm during the follow-up after AF ablation has proven to be more effective than traditional strategies, they continued. “The use of mobile ECG self-recording devices allows for earlier detection of AF recurrence and may empower patients to engage in shared health decision making.”