An intelligent model for classifying supraventricular tachycardia mechanisms based on 12-lead wearable electrocardiogram devices

doi:10.12122/j.issn.1673-4254.2024.05.06

Abstract

Abstract:

Objective To develop an intelligent model for differential diagnosis of atrioventricular nodal re-entrant tachycardia (AVNRT) and atrioventricular re-entrant tachycardia (AVRT) using 12-lead wearable electrocardiogram devices. Methods A total of 356 samples of 12-lead supraventricular tachycardia (SVT) electrocardiograms recorded by wearable devices were randomly divided into training and validation sets using 5-fold cross validation to establish the intelligent classification model, and 101 patients with the diagnosis of SVT undergoing electrophysiological studies and radiofrequency ablation from October, 2021 to March, 2023 were selected as the testing set. The changes in electrocardiogram parameters before and during induced tachycardia were compared. Based on multiscale deep neural network, an intelligent diagnosis model for classifying SVT mechanisms was constructed and validated. The 3-lead electrocardiogram signals from II, III, and V₁ were extracted to build new classification models, whose diagnostic efficacy was compared with that of the 12-lead model. Results Of the 101 patients with SVT in the testing set, 68 were diagnosed with AVNRT and 33 were diagnosed with AVRT by electrophysiological study. The pre-trained model achieved a high area under the precision-recall curve (0.9492) and F1 score (0.8195) for identifying AVNRT in the validation set. The total F1 scores of the lead II, III, V₁, 3-lead and 12-lead intelligent diagnostic models in the testing set were 0.5597, 0.6061, 0.3419, 0.6003 and 0.6136, respectively. Compared with the 12-lead classification model, the lead-III model had a net reclassification index improvement of -0.029 (P=0.878) and an integrated discrimination index improvement of -0.005 (P=0.965). Conclusion The intelligent diagnostic model based on multiscale deep neural network using wearable electrocardiogram devices has an acceptable accuracy for classifying SVT mechanisms.

Key words: wearable electrocardiogram, supraventricular tachycardia, 12-lead electrocardiogram, multiscale deep neural network, atrioventricular nodal re-entrant tachycardia, atrioventricular re-entrant tachycardia

Hongsen WANG, Lijie MI, Yue ZHANG, Lan GE, Jiewei LAI, Tao CHEN, Jian LI, Xiangmin SHI, Jiancheng XIU, Min TANG, Wei YANG, Jun GUO. An intelligent model for classifying supraventricular tachycardia mechanisms based on 12-lead wearable electrocardiogram devices[J]. Journal of Southern Medical University, 2024, 44(5): 851-858.

Figures/Tables 8

References 31

1	Brugada J, Katritsis DG, Arbelo E, et al. 2019 ESC Guidelines for the management of patients with supraventricular tachycardiaThe Task Force for the management of patients with supraventricular tachycardia of the European Society of Cardiology (ESC)[J]. Eur Heart J, 2020, 41(5): 655-720. DOI: 10.1093/eurheartj/ehz827
2	Peng G, Zei PC. Diagnosis and management of paroxysmal supraventricular tachycardia[J]. JAMA, 2024, 331(7): 601-10. DOI: 10.1001/jama.2024.0076
3	Al Mehairi M, Al Ghamdi SA, Dagriri K, et al. The importance of utilizing 24-h Holter monitoring as a non-invasive method of predicting the mechanism of supraventricular tachycardia[J]. J Saudi Heart Assoc, 2011, 23(4): 241-3. DOI: 10.1016/j.jsha.2011.05.002
4	Reed MJ, Grubb NR, Lang CC, et al. Multi-centre Randomised Controlled Trial of a Smartphone-based Event Recorder Alongside Standard Care Versus Standard Care for Patients Presenting to the Emergency Department with Palpitations and Pre-syncope: the IPED (Investigation of Palpitations in the ED) study[J]. E Clin Med, 2019, 8: 37-46. DOI: 10.1186/s13063-018-3098-1
5	Toyohara K, Kudo Y, Takeuchi D, et al. Smartwatch detection of undiagnosed palpitations in a juvenile[J]. Intern Med, 2022, 61(17): 2691-2. DOI: 10.2169/internalmedicine.8621-21
6	Tabing A, Harrell TE, Romero S, et al. Supraventricular tachycardia diagnosed by smartphone ECG[J]. BMJ Case Rep, 2017, 2017: bcr2016217197. DOI: 10.1136/bcr-2016-217197
7	Bogossian H, Iliodromitis K, Robl S. Jump-Phänomen zur SVT-Induktion und die Korrelation mit der EKG-Dokumentation über die Apple Watch™[J]. Herzschrittmachertherapie+Elektrophysiologie, 2020, 31(4): 426-9. DOI: 10.1007/s00399-020-00722-7
8	Kasai YH, Kasai J, Sekiguchi Y, et al. Apple Watch® facilitates single-session catheter ablation of coexisting atrioventricular nodal reentrant tachycardia and atrioventricular reentrant tachycardia[J]. Clin Case Rep, 2021, 9(8): e04702. DOI: 10.1002/ccr3.4702
9	Wu SJ, Li CH, Lin JC, et al. Detecting supraventricular tachycardia with smartwatches facilitates the decision for catheter ablation: a case series[J]. Pacing Clin Electrophysiol, 2022, 45(1): 157-9. DOI: 10.1111/pace.14388
10	Matsunaga K, Inoue T, Miyake Y, et al. Twelve-lead electrocardiography identifies the therapeutic target of paroxysmal supraventricular tachycardia[J]. J Arrhythm, 2019, 36(1): 211-3. DOI: 10.1002/joa3.12261
11	Penfold MP, Cannon BC, Wackel PL. Empiric slow pathway cryoablation in symptomatic children without documented supraventricular tachycardia[J]. Pediatr Cardiol, 2024, 45(4): 921-5. DOI: 10.1007/s00246-022-03065-x
12	Cook DA, Oh SY, Pusic MV. Accuracy of physicians' electrocardiogram interpretations: a systematic review and meta-analysis[J]. JAMA Intern Med, 2020, 180(11): 1461-71. DOI: 10.1001/jamainternmed.2020.3989
13	Kaneko I, Hayano J, Yuda E. How can gender be identified from heart rate data? Evaluation using ALLSTAR heart rate variability big data analysis[J]. BMC Res Notes, 2023, 16(1): 5. DOI: 10.1186/s13104-022-06270-2
14	Galloway CD, Valys AV, Shreibati JB, et al. Development and validation of a deep-learning model to screen for hyperkalemia from the electrocardiogram[J]. JAMA Cardiol, 2019, 4(5): 428-36. DOI: 10.1001/jamacardio.2019.0640
15	Siontis KC, Noseworthy PA, Attia ZI, et al. Artificial intelligence-enhanced electrocardiography in cardiovascular disease management[J]. Nat Rev Cardiol, 2021, 18(7): 465-78. DOI: 10.1038/s41569-020-00503-2
16	Sau A, Ibrahim S, Kramer DB, et al. Artificial intelligence-enabled electrocardiogram to distinguish atrioventricular re-entrant tachycardia from atrioventricular nodal re-entrant tachycardia[J]. Cardiovasc Digit Health J, 2023, 4(2): 60-7. DOI: 10.1016/j.cvdhj.2023.01.004
17	王泓森, 陈韬, 颜勇, 等. 十二导联穿戴式心电设备院前诊断室上性心动过速的回顾性研究[J]. 中华保健医学杂志, 2022, 24(2): 87-90.
18	Lai JW, Tan HX, Wang JL, et al. Practical intelligent diagnostic algorithm for wearable 12-lead ECG via self-supervised learning on large-scale dataset[J]. Nat Commun, 2023, 14(1): 3741. DOI: 10.1038/s41467-023-39472-8
19	赖杰伟, 陈韵岱, 韩宝石, 等. 基于DenseNet的心电数据自动诊断算法[J]. 南方医科大学学报, 2019, 39(1): 69-75.
20	Kucheryavskiy S, Rodionova O, Pomerantsev A. Procrustes cross-validation of multivariate regression models[J]. Anal Chim Acta, 2023, 1255: 341096. DOI: 10.1016/j.aca.2023.341096
21	Liberman L, Pass RH, Starc TJ. Optimal surface electrocardiogram lead for identification of the mechanism of supraventricular tachycardia in children[J]. Pediatr Emerg Care, 2008, 24(1): 28-30. DOI: 10.1097/pec.0b013e31815f3992
22	Hannun AY, Rajpurkar P, Haghpanahi M, et al. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network[J]. Nat Med, 2019, 25(1): 65-9. DOI: 10.1038/s41591-018-0268-3
23	Zhang Y, Liu S, He ZH, et al. A CNN model for cardiac arrhythmias classification based on individual ECG signals[J]. Cardiovasc Eng Technol, 2022, 13(4): 548-57. DOI: 10.1007/s13239-021-00599-8
24	Yao Y, Yu BS, Gong C, et al. Understanding how pretraining regularizes deep learning algorithms[J]. IEEE Trans Neural Netw Learn Syst, 2023, 34(9): 5828-40. DOI: 10.1109/tnnls.2021.3131377
25	Perlman O, Katz A, Amit G, et al. Supraventricular tachycardia classification in the 12-lead ECG using atrial waves detection and a clinically based tree scheme[J]. IEEE J Biomed Health Inform, 2016, 20(6): 1513-20. DOI: 10.1109/jbhi.2015.2478076
26	Attia ZI, Noseworthy PA, Lopez-Jimenez F, et al. An artificial intelligence-enabled ECG algorithm for the identification of patients with atrial fibrillation during sinus rhythm: a retrospective analysis of outcome prediction[J]. Lancet, 2019, 394(10201): 861-7. DOI: 10.1016/s0140-6736(19)31721-0
27	Jo YY, Kwon JM, Jeon KH, et al. Artificial intelligence to diagnose paroxysmal supraventricular tachycardia using electrocardiography during normal sinus rhythm[J]. Eur Heart J Digit Health, 2021, 2(2): 290-8. DOI: 10.1093/ehjdh/ztab025
28	Shen CP, Muse ED. Towards an artificial intelligence-augmented, ECG-enabled physical exam[J]. Lancet Digit Health, 2022, 4(2): e78-9. DOI: 10.1016/s2589-7500(21)00281-8
29	Senoner T, Pfeifer B, Barbieri F, et al. Identifying the location of an accessory pathway in pre-excitation syndromes using an artificial intelligence-based algorithm[J]. J Clin Med, 2021, 10(19): 4394. DOI: 10.3390/jcm10194394
30	Ding JW, Tang Y, Chang RH, et al. Reduction in the motion artifacts in noncontact ECG measurements using a novel designed electrode structure[J]. Sensors, 2023, 23(2): 956. DOI: 10.3390/s23020956
31	Boriani G, Svennberg E, Guerra F, et al. Reimbursement practices for use of digital devices in atrial fibrillation and other arrhythmias: a European Heart Rhythm Association survey[J]. Europace, 2022, 24(11): 1834-43. DOI: 10.1093/europace/euac142

Item	AVNRT (n=68)	AVRT (n=33)	Total (n=101)	P
Age (years)	47.68±14.24	42.09±13.72	45.85±14.25	0.064
BMI (kg/m²)	24.35±3.44	23.98±2.92	24.23±3.27	0.591
Male	28 (41.2%)	19 (57.6%)	47 (46.5%)	0.121
Heart rate (bpm)	78.97±13.29	76.48±11.02	78.16±12.60	0.355
SBP (mmHg)	131.25±14.33	127.97±15.77	130.18±14.81	0.299
DBP (mmHg)	78.28±10.85	75.82±11.88	77.48±11.20	0.302
K⁺(mmol/L)	3.84±0.29	4.01±0.33	3.89±0.32	0.010^†
Na⁺(mmol/L)	140.60±1.96	141.50±2.69	140.89±2.25	0.058
Cl^-(mmol/L)	104.50±2.17	105.45±2.12	104.81±2.19	0.040^†
Ca²⁺(mmol/L)	2.32±0.12	2.36±0.19	2.33±0.15	0.181
Palpitation	65 (95.6%)	33 (100.0%)	98 (97.0%)	0.549
Chest pain	25 (36.8%)	17 (51.5%)	42 (41.6%)	0.158
Weak	5 (7.4%)	3 (9.1%)	8 (7.9%)	0.714
Sweating	9 (13.2%)	7 (21.2%)	16 (15.8%)	0.303
Neck palpitation	3 (4.4%)	2 (6.1%)	5 (5.0%)	0.661
Pant	7 (10.3%)	4 (12.1%)	11 (10.9%)	0.746
Syncope	6 (8.8%)	3 (9.1%)	9 (8.9%)	1.000
Headache	10 (14.7%)	3 (9.1%)	13 (12.9%)	0.538
Age at first onset (years)	39.06±18.26	33.00±15.09	37.08±17.45	0.102
Number of episodes	18.21±27.56	40.82±90.68	25.59±57.04	0.170
Duration (min)				0.095
<10	9 (13.2%)	11 (33.3%)	20 (19.8%)
10-30	21 (30.9%)	6 (18.2%)	27 (26.7%)
30-120	22 (32.4%)	8 (24.2%)	30 (29.7%)
>120	16 (23.5%)	8 (24.2%)	24 (23.8%)
Smoking	10 (14.7%)	8 (24.2%)	18 (17.8%)	0.240
Drinking	8 (11.8%)	3 (9.1%)	11 (10.9%)	1.000
Beta blockers	3 (4.4%)	3 (9.1%)	6 (5.9%)	0.389
CCB	3 (4.4%)	3 (9.1%)	6 (5.9%)	0.389
Hypertension	15 (22.1%)	5 (15.2%)	20 (19.8%)	0.414
AF	2 (2.9%)	0 (0.0%)	2 (1.0%)	1.000
Diabetes	3 (4.5%)	0 (0.0%)	3 (3.0%)	0.549
CAD	3 (4.4%)	2 (6.1%)	5 (5.0%)	0.661

Item	AVNRT (n=68)	AVRT (n=33)	Total (n=101)	P
Age (years)	47.68±14.24	42.09±13.72	45.85±14.25	0.064
BMI (kg/m²)	24.35±3.44	23.98±2.92	24.23±3.27	0.591
Male	28 (41.2%)	19 (57.6%)	47 (46.5%)	0.121
Heart rate (bpm)	78.97±13.29	76.48±11.02	78.16±12.60	0.355
SBP (mmHg)	131.25±14.33	127.97±15.77	130.18±14.81	0.299
DBP (mmHg)	78.28±10.85	75.82±11.88	77.48±11.20	0.302
K⁺(mmol/L)	3.84±0.29	4.01±0.33	3.89±0.32	0.010^†
Na⁺(mmol/L)	140.60±1.96	141.50±2.69	140.89±2.25	0.058
Cl^-(mmol/L)	104.50±2.17	105.45±2.12	104.81±2.19	0.040^†
Ca²⁺(mmol/L)	2.32±0.12	2.36±0.19	2.33±0.15	0.181
Palpitation	65 (95.6%)	33 (100.0%)	98 (97.0%)	0.549
Chest pain	25 (36.8%)	17 (51.5%)	42 (41.6%)	0.158
Weak	5 (7.4%)	3 (9.1%)	8 (7.9%)	0.714
Sweating	9 (13.2%)	7 (21.2%)	16 (15.8%)	0.303
Neck palpitation	3 (4.4%)	2 (6.1%)	5 (5.0%)	0.661
Pant	7 (10.3%)	4 (12.1%)	11 (10.9%)	0.746
Syncope	6 (8.8%)	3 (9.1%)	9 (8.9%)	1.000
Headache	10 (14.7%)	3 (9.1%)	13 (12.9%)	0.538
Age at first onset (years)	39.06±18.26	33.00±15.09	37.08±17.45	0.102
Number of episodes	18.21±27.56	40.82±90.68	25.59±57.04	0.170
Duration (min)				0.095
<10	9 (13.2%)	11 (33.3%)	20 (19.8%)
10-30	21 (30.9%)	6 (18.2%)	27 (26.7%)
30-120	22 (32.4%)	8 (24.2%)	30 (29.7%)
>120	16 (23.5%)	8 (24.2%)	24 (23.8%)
Smoking	10 (14.7%)	8 (24.2%)	18 (17.8%)	0.240
Drinking	8 (11.8%)	3 (9.1%)	11 (10.9%)	1.000
Beta blockers	3 (4.4%)	3 (9.1%)	6 (5.9%)	0.389
CCB	3 (4.4%)	3 (9.1%)	6 (5.9%)	0.389
Hypertension	15 (22.1%)	5 (15.2%)	20 (19.8%)	0.414
AF	2 (2.9%)	0 (0.0%)	2 (1.0%)	1.000
Diabetes	3 (4.5%)	0 (0.0%)	3 (3.0%)	0.549
CAD	3 (4.4%)	2 (6.1%)	5 (5.0%)	0.661

Item	AVNRT (n=68)	AVRT (n=33)	Total (n=101)	P
PR intervals (ms)	145.47±18.31	147.12±34.13	146.02±24.56	0.796
QRS duration (ms)	92.08±12.18	105.91±22.59	96.69±17.55	0.002
QT (ms)	383.77±34.87	380.12±50.55	382.56±40.55	0.675
QTc (ms)	425.42±28.30	429.58±24.94	426.81±27.17	0.476

Item	AVNRT (n=68)	AVRT (n=33)	Total (n=101)	P
PR intervals (ms)	145.47±18.31	147.12±34.13	146.02±24.56	0.796
QRS duration (ms)	92.08±12.18	105.91±22.59	96.69±17.55	0.002
QT (ms)	383.77±34.87	380.12±50.55	382.56±40.55	0.675
QTc (ms)	425.42±28.30	429.58±24.94	426.81±27.17	0.476

Parameter	AVNRT (n=68)	AVRT (n=33)	Total (n=101)	P
HR (bpm)	169.46±29.81	161.52±21.71	166.86±27.57	0.176
QRS duration (ms)	77.19±10.22	79.30±15.31	77.88±12.08	0.413
QT (ms)	292.66±35.08	304.00±37.07	296.37±35.96	0.138
QTc (ms)	472.40±48.20	476.58±68.02	473.76±55.14	0.723
QTcd (ms)	143.74±80.20	170.06±91.56	152.34±84.53	0.143