Cardiac anomaly detection based on time and frequency domain features using tree-based classifiers.

OBJECTIVE: Recent advantages in mHealth-enabled ECG recorders boosted the demand for algorithms, which are able to automatically detect cardiac anomalies with high accuracy. APPROACH: We present a combined method of classical signal analysis and machine learning which has been developed during the Computing in Cardiology Challenge (CinC) 2017. Almost 400 hand-crafted features have been developed to reflect the complex physiology of cardiac arrhythmias and their appearance in single-channel ECG recordings. For the scope of this article, we performed several experiments on the publicly available challenge dataset to improve the classification accuracy. We compared the performance of two tree-based algorithms-gradient boosted trees and random forests-using different parameters for learning. We assessed the influence of five different sets of training annotations on the classifiers performance. Further, we present a new web-based ECG viewer to review and correct the training labels of a signal data set. Moreover, we analysed the feature importance and evaluated the model performance when using only a subset of the features. The primary data source used in the analysis was the dataset of the CinC 2017, consisting of 8528 signals from four classes. Our best results were achieved using a gradient boosted tree model which worked significantly better than random forests. MAIN RESULTS: Official results of the challenge follow-up phase provided by the Challenge organizers on the full hidden test set are 90.8% (Normal), 84.1% (AF), 74.5% (Other), resulting in a mean F1-score of 83.2%, which was only 1.6% behind the challenge winner and 0.2% ahead of the next-best algorithm. Official results were rounded to two decimal places which lead to the equal-second best F1 F -score of 83% with five others. SIGNIFICANCE: The algorithm achieved the second-best score among 80 algorithms of the Challenge follow-up phase equal with five others.

Near Field Communication-based telemonitoring with integrated ECG recordings.

OBJECTIVES: Telemonitoring of vital signs is an established option in treatment of patients with chronic heart failure (CHF). In order to allow for early detection of atrial fibrillation (AF) which is highly prevalent in the CHF population telemonitoring programs should include electrocardiogram (ECG) signals. It was therefore the aim to extend our current home monitoring system based on mobile phones and Near Field Communication technology (NFC) to enable patients acquiring their ECG signals autonomously in an easy-to-use way. METHODS: We prototypically developed a sensing device for the concurrent acquisition of blood pressure and ECG signals. The design of the device equipped with NFC technology and Bluetooth allowed for intuitive interaction with a mobile phone based patient terminal. This ECG monitoring system was evaluated in the course of a clinical pilot trial to assess the system's technical feasibility, usability and patient's adherence to twice daily usage. RESULTS: 21 patients (4f, 54 ± 14 years) suffering from CHF were included in the study and were asked to transmit two ECG recordings per day via the telemonitoring system autonomously over a monitoring period of seven days. One patient dropped out from the study. 211 data sets were transmitted over a cumulative monitoring period of 140 days (overall adherence rate 82.2%). 55% and 8% of the transmitted ECG signals were sufficient for ventricular and atrial rhythm assessment, respectively. CONCLUSIONS: Although ECG signal quality has to be improved for better AF detection the developed communication design of joining Bluetooth and NFC technology in our telemonitoring system allows for ambulatory ECG acquisition with high adherence rates and system usability in heart failure patients.

Initial experiences with a telemedicine framework for remote pacemaker follow-up.

According to international guidelines implanted cardiac pacemakers (PM) have to be checked periodically to ensure that they are working correctly. To spare a significant number of patients the burden of traveling to specialized PM clinics a telemedicine framework has been developed prototypically. A mobile, personal digital assistant (PDA) based PM follow-up unit provides the caregiver at the point-of-care with the necessary infrastructure to perform a basic PM follow-up examination remotely. In case of detected malfunction of the PM the patient is ordered to the hospital for further examination. The system has been evaluated in a clinical pilot trial on 44 patients with a total of 23 different PM models from 8 different manufacturers. The initial results indicate the potential of the concept to work as an efficient, manufacturer independent screening method with the ultimate goal to increase the safety, quality and efficiency of PM therapy.

Automated and manufacturer independent assessment of the battery status of implanted cardiac pacemakers by electrocardiogram analysis.

According to international standards, cardiac pacemakers have to indicate the status of their batteries upon magnet application by specific stimulation patterns. The purpose of this study has been to assess whether this concept can be used as a basis for automated and manufacturer independent examination of the depletion level of pacemakers in the framework of a collaborative telemedical pacemaker follow-up system. A prototype of such a system was developed and tested in a real clinical environment. Electrocardiograms (ECGs) were recorded during magnet application and automatically processed to extract the specific stimulation patterns. The results were used to assign each signal a corresponding pacemaker status: "ok," "replace" or "undefined," based on the expected behavior of the devices as specified by the manufacturer. The outcome of this procedure was compared to the result of an expert examination, resulting in a positive predictive value of 100% for the detection of ECGs indicating pacemaker status "ok." The method can, therefore, be utilized to quickly, safely and manufacturer neutrally classify cases into the categories "ok" and "needs further checking," which - in a telemedical setting - may be used to increase the efficiency of pacemaker follow-up procedures in the future.