The clinical challenge is both vast and elusive. Atrial fibrillation (AF), a chaotic quivering of the upper chambers of the heart, affects tens of millions of people globally and multiplies the risk of ischemic stroke fivefold. Yet in its paroxysmal form, AF can appear for minutes, vanish for weeks, and remain entirely silent. A resting ECG in a physician’s office captures roughly ten seconds of electrical activity. Even a Holter monitor, typically worn for 24 to 48 hours, often fails to intersect with these sporadic episodes. The consequence is a diagnostic gap through which high-risk patients slip, sometimes until a second, devastating stroke occurs. Professor Vaidotas Marozas, Director of the Biomedical Engineering Institute at KTU, explains that the team deliberately focused on post-stroke patients because in this group, missing a brief run of AF can mean withholding anticoagulation therapy and accepting a preventable recurrence.

The wristband’s first line of detection is a photoplethysmography (PPG) sensor that continuously tracks blood volume changes in the microvascular bed of the skin. From this optical signal, the device extracts pulse intervals and feeds them into a real-time arrhythmia detection algorithm. When the algorithm identifies an irregularity suggestive of AF, the wristband vibrates and displays a notification, inviting the user to initiate a more detailed recording. This is where the technology departs from a simple fitness tracker. The wristband incorporates multiple dry electrodes—some on the back in contact with the skin, others on the bezel that the opposite hand touches—forming a closed circuit across the chest. Within about sixty seconds, the device synthesizes a multi-lead ECG that captures the heart’s electrical vector from several angles, producing a tracing that approaches the diagnostic quality of a standard 12-lead recording.

The raw signals, however, arrive in a hostile environment. Everyday motion, muscle tension, and shifting skin contact generate artifacts that can swamp the low-amplitude cardiac waveforms. To address this, the KTU team embedded a multi-stage signal quality assessment algorithm. Before any strip is stored or transmitted, the system classifies segments as usable or corrupted, discarding noisy data and retaining only those intervals where the signal-to-noise ratio crosses a clinically meaningful threshold. The clean ECG segments are then sent to the cloud via a secure cellular link, allowing a cardiologist to review the rhythm remotely. By shifting the burden of artifact rejection onto the device itself, the system spares the clinician from sifting through hours of unreadable noise and dramatically reduces the false-alert rate that has plagued earlier ambulatory monitors.

Beyond simple detection, the patent covers a novel arrhythmia aggregation parameter that moves the diagnostic conversation from a binary “yes/no” to a temporal pattern. Instead of merely flagging that an arrhythmia occurred, the algorithm analyzes how episodes are distributed over the monitoring period. It calculates whether events are uniformly scattered or clustered into dense bursts separated by quiet intervals. Marozas likens this to measuring not just the total rainfall but the sequence of storms. For a physician, this aggregation metric can reveal whether a patient’s paroxysmal AF is becoming more frequent or organizing into longer runs, signaling a progression in the underlying atrial substrate and a mounting stroke risk long before the symptoms become obvious. Such a parameter could eventually guide decisions about when to initiate, adjust, or pause anticoagulation therapy.

The technical backbone of the wristband is the product of an unusually interdisciplinary collaboration. The patent lists KTU researcher Andrius Petrėnas, a specialist in biomedical signal processing, alongside cardiologist Justinas Bacevičius from Vilnius University Hospital Santaros Clinics, and engineers Andrius Sološenko, Saulius Daukantas, and Monika Butkuvienė. Medical residents from the same hospital contributed clinical validation data, while KTU doctoral students refined the machine learning models that underpin the beat classification. The team’s recent work has focused on hardening the algorithms against real-world variability—different skin tones, ambient temperatures, and movement patterns—to ensure the device performs reliably not only in a controlled lab but in kitchens, stairwells, and crowded streets.

The commercial translation of the research into the TeltoHeart wristband, manufactured by Teltonika Telemedic, means that the technology is already available in Lithuania through telemedicine projects funded by the European Union. Patients recovering at home after surgery or a serious illness can now send their cardiac data to a specialist without the travel and wait times that burden traditional outpatient clinics. The wristband is intended to be used in concert with a physician’s consultation, not as a standalone diagnostic tool, and the company is working to integrate the data stream into existing electronic health record systems. The immediate goal is not to replace the cardiologist but to extend the clinic’s reach into the patient’s daily life, turning the weeks between appointments into a continuous, intelligently filtered diagnostic window.

Broader reimbursement for such continuous remote monitoring, particularly for high-risk populations like stroke survivors, remains a future objective. The developers acknowledge that the ultimate test of the technology will be a large-scale clinical trial demonstrating that the aggregation-guided approach leads to measurable reductions in recurrent stroke rates. Should that evidence materialize, the patented wristband and its algorithms could reshape the standard of care for paroxysmal atrial fibrillation, transforming a frustrating game of diagnostic chance into a proactive, data-rich surveillance strategy. For now, the patent stands as recognition of a decade of signal processing research and a crucial bridge from laboratory prototypes to a device that quietly, persistently, watches over the heart.

Subject of Research: Continuous photoplethysmography-based arrhythmia monitoring with prompted multi-lead ECG and arrhythmia aggregation analysis for detection of paroxysmal atrial fibrillation in post-stroke patients.
Article Title: A Wristband That Hunts the Hidden Arrhythmias Behind Stroke
News Publication Date: [Not available]
Web References: [Not available]
References: [Not available]
Image Credits: KTU
Keywords: atrial fibrillation, stroke, photoplethysmography, wearable ECG, arrhythmia aggregation, remote monitoring, digital health, patent, biomedical engineering