Searching for non-sense: identification of pacemaker non-sense and non-capture failures using machine learning techniques

Abnormal or unexpected function of pacemakers due to mechanical failure of the implantation, electrical failures of the battery and electrodes, or physiological failures to respond to the stimulus may cause harm to a patient. A novel Bayesian decision tree algorithm is proposed to detect two types of pacemaker failures, non-sense and non-capture, without a priori knowledge of pacemaker type, model, or programming. A variety of pacemaker devices and modes were studied, including devices with single and dual chamber pacing; single and dual chamber sensing; and fixed rate and rate adaptive pacing. 12-lead ECG signals were acquired from 34 pacemaker patients at rest. These signals were annotated by a team of experts. A 10-fold cross-validation was performed on the data set to test the algorithm. Out-of-sample sensitivity and specificity of 87.8% and 98.7%, respectively, were achieved. This work shows that non-sense and non-captures pacemaker failures can be detected with high sensitivity and specificity without prior knowledge of the pacemaker type, model or programming, making this algorithm clinically relevant in emergency room environments where such pacemaker information may be unavailable.