Predicting troponin biomarker elevation from electrocardiograms using a deep neural network
Background Elevated troponin levels are a sensitive biomarker for cardiac injury. The quick and reliable prediction of troponin elevation for patients with chest pain from readily available ECGs may pose a valuable time-saving diagnostic tool during decision-making concerning this patient population...
Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMJ Publishing Group
2024-10-01
|
| Series: | Open Heart |
| Online Access: | https://openheart.bmj.com/content/11/2/e002937.full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850284896054935552 |
|---|---|
| author | Truls Råmunddal Erik Andersson Zacharias Mandalenakis Kristofer Skoglund Araz Rawshani Elmir Omerovic Jan Borén Petur Petursson Aidin Rawshani Peter Lundgren Christian Dworeck Lukas Hilgendorf Vibha Gupta Charlotta Ljungman Gustav Smith |
| author_facet | Truls Råmunddal Erik Andersson Zacharias Mandalenakis Kristofer Skoglund Araz Rawshani Elmir Omerovic Jan Borén Petur Petursson Aidin Rawshani Peter Lundgren Christian Dworeck Lukas Hilgendorf Vibha Gupta Charlotta Ljungman Gustav Smith |
| author_sort | Truls Råmunddal |
| collection | DOAJ |
| description | Background Elevated troponin levels are a sensitive biomarker for cardiac injury. The quick and reliable prediction of troponin elevation for patients with chest pain from readily available ECGs may pose a valuable time-saving diagnostic tool during decision-making concerning this patient population.Methods and results The data used included 15 856 ECGs from patients presenting to the emergency rooms with chest pain or dyspnoea at two centres in Sweden from 2015 to June 2023. All patients had high-sensitivity troponin test results within 6 hours after 12-lead ECG. Both troponin I (TnI) and TnT were used, with biomarker-specific cut-offs and sex-specific cut-offs for TnI. On this dataset, a residual convolutional neural network (ResNet) was trained 10 times, each on a unique split of the data. The final model achieved an average area under the curve for the receiver operating characteristic curve of 0.7717 (95% CI±0.0052), calibration curve analysis revealed a mean slope of 1.243 (95% CI±0.075) and intercept of −0.073 (95% CI±0.034), indicating a good correlation between prediction and ground truth. Post-classification, tuned for F1 score, accuracy was 71.43% (95% CI±1.28), with an F1 score of 0.5642 (95% CI±0.0052) and a negative predictive value of 0.8660 (95% CI±0.0048), respectively. The ResNet displayed comparable or surpassing metrics to prior presented models.Conclusion The model exhibited clinically meaningful performance, notably its high negative predictive accuracy. Therefore, clinical use of comparable neural networks in first-line, quick-response triage of patients with chest pain or dyspnoea appears as a valuable option in future medical practice. |
| format | Article |
| id | doaj-art-7f50cac17cc74fe38d72da0f369b5ec8 |
| institution | OA Journals |
| issn | 2053-3624 |
| language | English |
| publishDate | 2024-10-01 |
| publisher | BMJ Publishing Group |
| record_format | Article |
| series | Open Heart |
| spelling | doaj-art-7f50cac17cc74fe38d72da0f369b5ec82025-08-20T01:47:26ZengBMJ Publishing GroupOpen Heart2053-36242024-10-0111210.1136/openhrt-2024-002937Predicting troponin biomarker elevation from electrocardiograms using a deep neural networkTruls Råmunddal0Erik Andersson1Zacharias Mandalenakis2Kristofer Skoglund3Araz Rawshani4Elmir Omerovic5Jan Borén6Petur Petursson7Aidin Rawshani8Peter Lundgren9Christian Dworeck10Lukas Hilgendorf11Vibha Gupta12Charlotta Ljungman13Gustav Smith14Department of Cardiology, Sahlgrenska University Hospital, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Cardiology, Sahlgrenska University Hospital, Goteborg, SwedenDepartment of Cardiology, Sahlgrenska University Hospital, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg, Goteborg, SwedenDepartment of Cardiology, Sahlgrenska University Hospital, Goteborg, SwedenBackground Elevated troponin levels are a sensitive biomarker for cardiac injury. The quick and reliable prediction of troponin elevation for patients with chest pain from readily available ECGs may pose a valuable time-saving diagnostic tool during decision-making concerning this patient population.Methods and results The data used included 15 856 ECGs from patients presenting to the emergency rooms with chest pain or dyspnoea at two centres in Sweden from 2015 to June 2023. All patients had high-sensitivity troponin test results within 6 hours after 12-lead ECG. Both troponin I (TnI) and TnT were used, with biomarker-specific cut-offs and sex-specific cut-offs for TnI. On this dataset, a residual convolutional neural network (ResNet) was trained 10 times, each on a unique split of the data. The final model achieved an average area under the curve for the receiver operating characteristic curve of 0.7717 (95% CI±0.0052), calibration curve analysis revealed a mean slope of 1.243 (95% CI±0.075) and intercept of −0.073 (95% CI±0.034), indicating a good correlation between prediction and ground truth. Post-classification, tuned for F1 score, accuracy was 71.43% (95% CI±1.28), with an F1 score of 0.5642 (95% CI±0.0052) and a negative predictive value of 0.8660 (95% CI±0.0048), respectively. The ResNet displayed comparable or surpassing metrics to prior presented models.Conclusion The model exhibited clinically meaningful performance, notably its high negative predictive accuracy. Therefore, clinical use of comparable neural networks in first-line, quick-response triage of patients with chest pain or dyspnoea appears as a valuable option in future medical practice.https://openheart.bmj.com/content/11/2/e002937.full |
| spellingShingle | Truls Råmunddal Erik Andersson Zacharias Mandalenakis Kristofer Skoglund Araz Rawshani Elmir Omerovic Jan Borén Petur Petursson Aidin Rawshani Peter Lundgren Christian Dworeck Lukas Hilgendorf Vibha Gupta Charlotta Ljungman Gustav Smith Predicting troponin biomarker elevation from electrocardiograms using a deep neural network Open Heart |
| title | Predicting troponin biomarker elevation from electrocardiograms using a deep neural network |
| title_full | Predicting troponin biomarker elevation from electrocardiograms using a deep neural network |
| title_fullStr | Predicting troponin biomarker elevation from electrocardiograms using a deep neural network |
| title_full_unstemmed | Predicting troponin biomarker elevation from electrocardiograms using a deep neural network |
| title_short | Predicting troponin biomarker elevation from electrocardiograms using a deep neural network |
| title_sort | predicting troponin biomarker elevation from electrocardiograms using a deep neural network |
| url | https://openheart.bmj.com/content/11/2/e002937.full |
| work_keys_str_mv | AT trulsramunddal predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT erikandersson predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT zachariasmandalenakis predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT kristoferskoglund predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT arazrawshani predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT elmiromerovic predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT janboren predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT peturpetursson predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT aidinrawshani predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT peterlundgren predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT christiandworeck predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT lukashilgendorf predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT vibhagupta predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT charlottaljungman predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork AT gustavsmith predictingtroponinbiomarkerelevationfromelectrocardiogramsusingadeepneuralnetwork |