Low Workload Concept for the Detection of Silent Atrial Fibrillation (AF) and Atrial Fibrillation Burden in Patients At High Risk of AF and Stroke

Overview

Patient-centered novel e-health technology and services will lay the foundation for future healthcare systems and services to support health and welfare promotion. Yet, there is a lack of ways to incorporate novel technological innovations into easy-to-use, cost-effective and low workload treatment. The detection of atrial fibrillation (AF) paroxysms and its permanent form as well as the prevention of AF-related strokes are major challenges in cardiology today. AF is often silent or asymptomatic, but the risk of ischemic stroke seems to be similar regardless of the presence or absence of symptoms. CARE-DETECT algorithm development part I will investigate following topics: 1. The usefulness and validity of bed sensor and mobile phone application in rhythm disorder capture compared to gold standard ECG-holter monitoring (Faros ECG) 2. Accuracy of AF detection from PDL data 3. Technical development of algorithms to detect arrhythmia from data collected with these novel devices 4. Development of a pre-processing tool that will evaluate the collected data and generate a preliminary filtered report of the raw data to ease clinician's workload in data handling and rhythm evaluation. CARE-DETECT clinical trial (part II) proposal provides a new concept for low workload for healthcare personnel, high diagnostic yield in silent AF detection and AF burden evaluation. CARE-DETECT protocol proposal seeks to address following issues: 1. Can a combination of actively used smartphone application and passive monitoring with bed sensor (with upstream ECG) - compared to routine care - enhance the detection of AF in patients who are at increased risk of stroke and have undergone a cardiac procedure? 2. What is the actual AF burden in paroxysmal AF patients after the detection of new-onset AF? 3. Can a direct-to-consumer telehealth with integrated cloud-based telecardiology service for medical professionals improve the efficacy of silent AF detection and what is the AF burden in patients suffering of (asymptomatic) paroxysmal AF and secondarily what is the cost-effectiveness of these new screening methods? 4. Additionally, during the hospitalization phase of the study part II PDL data will be collected in the intervention group. PDL data will be analyzed offline with the purpose to develop new methods and will not be used to monitor treatment or for diagnosis.

The detection of atrial fibrillation (AF) paroxysms and its permanent form as well as the prevention of AF-related strokes are major challenges in cardiology today. AF is often silent or asymptomatic, but the risk of ischemic stroke seems to be similar regardless of the presence or absence of symptoms. Asymptomatic AF patients are more likely to evade diagnostic effort and without appropriate anticoagulation they are left vulnerable to thromboembolism and ischemic stroke. Approximately one third of all ischemic strokes are of an unknown cause. Recent studies have shown that more diligent monitoring of heart rhythm with ambulatory devices after a cryptogenic stroke uncovers a high number of silent AF episodes. Timely detection of silent AF is challenging and, therefore, a stroke is still too often the first clinical manifestation of AF accounting for 22% of AF-related strokes. In order to detect symptoms appearing at periodic or random intervals, a capability for longer-term monitoring, e.g. for several days or weeks at a time, is required. Implantable electrocardiogram (ECG) loop recorders perform well in AF detection, but they are hardly feasible in large patient cohorts due to invasive nature and costs. Thus, there is an unmet clinical need for better AF detection tools. During the Automated Detection of Atrial Fibrillation via a Miniature Accelerometer and Gyroscope (NoStroke) project (as well as other projects carried out simultaneously at Department of Future Technologies, University of Turku, mainly funded by Tekes) a smartphone application has been developed for the detection of AF, which is implemented as a stand-alone solution to a common smartphone. The application utilizes the inbuilt motion sensors of the smartphones without any need to external sensors or equipment, such as electrodes or wires. This application (CardioSignal app) is a CE-marked medical device and will be used in this project and it will provide a venue for a clinical showcase of the application and more importantly, validate whether the application would be beneficial in the clinical settings in this follow-up study. Bed sensor (Emfit ltd) is a CE-marked device. It is based on ballistography method where a heartbeat-induced recoil can be measured while the subject is lying on the bed. Electromechanical membrane sensor recognizes and registers the recoil, and subsequently, this data can be used to assess heart rate and other relevant vital function. Philips Data Logger (PDL) measures continuously photoplethysmogram (PPG) and accelerometer data from the top side of the wrist using green wavelength Light Emitting Diode (LED) and 3-axis accelerometer. Faros Holter-monitor measures single-lead ECG signal from the chest and is attached via FastFix electrode. The part I trial is open, prospective interventional trial. These recordings are used merely to technical testing, validity evaluation and algorithm development purposes and the data is not used in any clinical evaluation whatsoever. The part II is a prospective, randomized, open-label, interventional study. Upon receipt of the signed written informed consent and satisfactory documentation that the patient met all inclusion and had no exclusion criteria, the study subjects will be randomized to either intervention or control arm. The randomization allocation will take into account the procedure subgroup of the patient (open-heart surgery, aortic valve replacement and transcatheter aortic valve implantation (TAVI) or coronary bypass and percutaneous coronary intervention (PCI)). In both subgroups the randomization will be done 1:1. Patients will be randomized before hospital discharge. The number of study subjects needed was estimated to be 300. Complying with the original plan, interim analysis was performed when 100 cases were enrolled. Based on the interim analysis results, the target enrollment number of patients was reduced to 150 patients due to the high number of false positive alerts leading to high-workload protocol inadequate for clinical use as such.