Neural Network based -Arrhythmia Monitoring Device: APivotal Clinical Trial
Ardian Rizal1, Puspa Lestari2, Satria Mandala3, Budi Satrijo4, Sasmojo Widito5
1Ardian Rizal, Department of Cardiology and Vascular Medicine, Faculty of Medicine, Universitas Brawijaya.
2Puspa Lestari, Department of Cardiology and Vascular Medicine, Faculty of Medicine, Universitas Brawijaya.
3Satria Mandala, Telkom University.
4Budi Satrijo, Department of Cardiology and Vascular Medicine, Faculty of Medicine, Universitas Brawijaya.
5Sasmojo Widito, Department of Cardiology and Vascular Medicine, Faculty of Medicine, Universitas Brawijaya.
Manuscript received on 11 January 2020 | Revised Manuscript received on 07 February 2020 | Manuscript Published on 20 February 2020 | PP: 346-349 | Volume-9 Issue-3S January 2020 | Retrieval Number: C10220193S20/2020©BEIESP | DOI: 10.35940/ijitee.C1022.0193S20
Open Access | Editorial and Publishing Policies | Cite | Zenodo | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Arrhythmia or irregular heart beat had wide range of clinical manifestations, from benign arrhythmia that not need any medication to life threatening condition. It can occur permanently or intermittently. Intermittent arrhythmia needs specific diagnostic tools that can record the electrocardiogram continuously. This research was sought to analysed the sensitivity, specificity, Positive Predictive Value, and Negative Predictive Value of arrhythmia monitoring device that based on neural network based artificial intelligent. The pivotal clinical trial was involved a total 103 people (health and stable arrhythmia patients). This research used a diagnostic test by comparing the electrocardiography (ECG) from prototype with standard ECG for diagnose arrhythmia. The Arrhythmia Monitoring System that we developed has three hardware components; smartphones, server for arrhythmia detection and patchable ECG recorder. All three components are connected with internet of things (IoT) technology. The architecture of Arrhythmia software monitoring included ECG signals pre-processing, beats detection, features extraction for detecting VT/VF, and classification for detecting VT/VF. Features extraction such as heart rate variability (HRV) and T wave alternans. We compared the ECG of arrhythmia prototype monitoring device with standard Holter monitoring. We enrolled 103 patients. There was no significant difference of heart rate between arrhythmia prototype monitoring device and standard Holter (87.26 ± 11.2 vs 86.07±9.15, P=0.43). There was significant different of maximum and minimum heart rate between arrhythmia prototype monitoring device and standard holter monitoring (121.3±31.7 vs 131.0±10.8, p= 0.000, and 65.1±13.5 vs 73.07±10.02, p=0.000). This device has low sensitivity 80% (95% CI 75% – 82%) and high specificity 91.8% (95% CI 85% – 92%) for detecting the abnormal ECG. The Positive Predictive Value (PPV) was 63.2% (95% CI 58.8% – 67.52%) and Negative Predictive Value (NVP) was 96.3% (95% CI 94.7% – 98.3%). This device demonstrates an ability to detect PVC and PAC (Sensitivity 71.4% (95% CI 66.4% – 76.4%) and 75% (95% CI 72%-78%), Specificity 97.8% (95% CI 95.8-99.8%) and 91.7% (95% CI 83.4%- 99.7%, respectively). The PPV of this device to detect PVC and PAC was 71.4% (95% CI 66.4%-76.4%) and 72.7%. (95% CI 68.7%-76.7%) The NPV of this device to detect PVC and PAC was 97.8% (95% CI 95.8%-99.85) and 98.9% (95% CI 98.1%-99.7%), respectively. This study found that the device to be a valuable diagnostic tool that has relatively low sensitivity but high specificity for diagnosing Abnormal ECG, PVC and PAC. According to the results of our study, we found that the device to be a valuable diagnostic tool that has relatively high sensitivity and specificity for diagnosing Abnormal ECG, PVC and PAC.
Keywords: Arrhythmia; Arrhythmia Monitoring Device; Holter Monitoring.
Scope of the Article: Neural Information Processing