| 1. |
- Lima, Emilly M., et al.
(author)
-
Deep neural network-estimated electrocardiographic age as a mortality predictor
- 2021
-
In: Nature Communications. - : Springer Nature. - 2041-1723. ; 12:1
-
Journal article (peer-reviewed)abstract
- The electrocardiogram (ECG) is the most commonly used exam for the screening and evaluation of cardiovascular diseases. Here, the authors propose that the age predicted by artificial intelligence from the raw ECG tracing can be a measure of cardiovascular health and provide prognostic information. The electrocardiogram (ECG) is the most commonly used exam for the evaluation of cardiovascular diseases. Here we propose that the age predicted by artificial intelligence (AI) from the raw ECG (ECG-age) can be a measure of cardiovascular health. A deep neural network is trained to predict a patient's age from the 12-lead ECG in the CODE study cohort (n = 1,558,415 patients). On a 15% hold-out split, patients with ECG-age more than 8 years greater than the chronological age have a higher mortality rate (hazard ratio (HR) 1.79, p < 0.001), whereas those with ECG-age more than 8 years smaller, have a lower mortality rate (HR 0.78, p < 0.001). Similar results are obtained in the external cohorts ELSA-Brasil (n = 14,236) and SaMi-Trop (n = 1,631). Moreover, even for apparent normal ECGs, the predicted ECG-age gap from the chronological age remains a statistically significant risk predictor. These results show that the AI-enabled analysis of the ECG can add prognostic information.
|
|
| 2. |
- Paixão, Gabriela M. M., et al.
(author)
-
Electrocardiographic Predictors of Mortality: Data from a Primary Care Tele-Electrocardiography Cohort of Brazilian Patients
- 2021
-
In: Hearts. - : MDPI AG. - 2673-3846. ; 2:4, s. 449-458
-
Journal article (peer-reviewed)abstract
- Computerized electrocardiography (ECG) has been widely used and allows linkage to electronic medical records. The present study describes the development and clinical applications of an electronic cohort derived from a digital ECG database obtained by the Telehealth Network of Minas Gerais, Brazil, for the period 2010–2017, linked to the mortality data from the national information system, the Clinical Outcomes in Digital Electrocardiography (CODE) dataset. From 2,470,424 ECGs, 1,773,689 patients were identified. A total of 1,666,778 (94%) underwent a valid ECG recording for the period 2010 to 2017, with 1,558,421 patients over 16 years old; 40.2% were men, with a mean age of 51.7 [SD 17.6] years. During a mean follow-up of 3.7 years, the mortality rate was 3.3%. ECG abnormalities assessed were: atrial fibrillation (AF), right bundle branch block (RBBB), left bundle branch block (LBBB), atrioventricular block (AVB), and ventricular pre-excitation. Most ECG abnormalities (AF: Hazard ratio [HR] 2.10; 95% CI 2.03–2.17; RBBB: HR 1.32; 95%CI 1.27–1.36; LBBB: HR 1.69; 95% CI 1.62–1.76; first degree AVB: Relative survival [RS]: 0.76; 95% CI0.71–0.81; 2:1 AVB: RS 0.21 95% CI0.09–0.52; and RS 0.36; third degree AVB: 95% CI 0.26–0.49) were predictors of overall mortality, except for ventricular pre-excitation (HR 1.41; 95% CI 0.56–3.57) and Mobitz I AVB (RS 0.65; 95% CI 0.34–1.24). In conclusion, a large ECG database established by a telehealth network can be a useful tool for facilitating new advances in the fields of digital electrocardiography, clinical cardiology and cardiovascular epidemiology.
|
|
| 3. |
- Ribeiro, Antônio H., et al.
(author)
-
Automatic 12-lead ECG Classification Using a Convolutional Network Ensemble
- 2020
-
In: 2020 Computing in Cardiology. - Rimini, Italy : IEEE. - 9781728173825 - 9781728111056
-
Conference paper (peer-reviewed)abstract
- The 12-lead electrocardiogram (ECG) is a major diagnostic test for cardiovascular diseases and enhanced automated analysis tools might lead to more reliable diagnosis and improved clinical practice. Deep neural networksare models composed of stacked transformations that learntasks by examples. Inspired by the success of these modelsin computer vision, we propose an end-to-end approach forthe task at hand. We trained deep convolutional neural network models in the heterogeneous dataset provided in thePhysionet 2020 Challenge and used an ensemble of sevenof these convolutional models for the classification of abnormalities present in the ECG records. Ensembles use theoutput of multiple models to generate a combined prediction and are known to improve performance and generalization when compared to the individual models. In oursubmission, we use an ensemble of neural networks withthe architecture similar to the one described in Nat Commun 11, 1760 (2020) for 12-lead ECGs classification. Ourapproach achieved a challenge validation score of 0.657,and full test score of 0.132, placing us, the “Code Team”,in 28 out of 41 in the official ranking.
|
|
| 4. |
- Ribeiro, Antônio H., et al.
(author)
-
Automatic diagnosis of the 12-lead ECG using a deep neural network
- 2020
-
In: Nature Communications. - : NATURE PUBLISHING GROUP. - 2041-1723. ; 11:1
-
Journal article (peer-reviewed)abstract
- The role of automatic electrocardiogram (ECG) analysis in clinical practice is limited by the accuracy of existing models. Deep Neural Networks (DNNs) are models composed of stacked transformations that learn tasks by examples. This technology has recently achieved striking success in a variety of task and there are great expectations on how it might improve clinical practice. Here we present a DNN model trained in a dataset with more than 2 million labeled exams analyzed by the Telehealth Network of Minas Gerais and collected under the scope of the CODE (Clinical Outcomes in Digital Electrocardiology) study. The DNN outperform cardiology resident medical doctors in recognizing 6 types of abnormalities in 12-lead ECG recordings, with F1 scores above 80% and specificity over 99%. These results indicate ECG analysis based on DNNs, previously studied in a single-lead setup, generalizes well to 12-lead exams, taking the technology closer to the standard clinical practice. The role of automatic electrocardiogram (ECG) analysis in clinical practice is limited by the accuracy of existing models. In that context, the authors present a Deep Neural Network (DNN) that recognizes different abnormalities in ECG recordings which matches or outperform cardiology and emergency resident medical doctors.
|
|
| 5. |
- Bentham, James, et al.
(author)
-
A century of trends in adult human height
- 2016
-
In: eLIFE. - 2050-084X. ; 5
-
Journal article (peer-reviewed)abstract
- Being taller is associated with enhanced longevity, and higher education and earnings. We reanalysed 1472 population-based studies, with measurement of height on more than 18.6 million participants to estimate mean height for people born between 1896 and 1996 in 200 countries. The largest gain in adult height over the past century has occurred in South Korean women and Iranian men, who became 20.2 cm (95% credible interval 17.522.7) and 16.5 cm (13.319.7) taller, respectively. In contrast, there was little change in adult height in some sub-Saharan African countries and in South Asia over the century of analysis. The tallest people over these 100 years are men born in the Netherlands in the last quarter of 20th century, whose average heights surpassed 182.5 cm, and the shortest were women born in Guatemala in 1896 (140.3 cm; 135.8144.8). The height differential between the tallest and shortest populations was 19-20 cm a century ago, and has remained the same for women and increased for men a century later despite substantial changes in the ranking of countries.
|
|
| 6. |
- Bentham, James, et al.
(author)
-
A century of trends in adult human height
- 2016
-
In: eLIFE. - : eLife Sciences Publications Ltd. - 2050-084X. ; 5
-
Journal article (peer-reviewed)abstract
- Being taller is associated with enhanced longevity, and higher education and earnings. We reanalysed 1472 population-based studies, with measurement of height on more than 18.6 million participants to estimate mean height for people born between 1896 and 1996 in 200 countries. The largest gain in adult height over the past century has occurred in South Korean women and Iranian men, who became 20.2 cm (95% credible interval 17.5–22.7) and 16.5 cm (13.3– 19.7) taller, respectively. In contrast, there was little change in adult height in some sub-Saharan African countries and in South Asia over the century of analysis. The tallest people over these 100 years are men born in the Netherlands in the last quarter of 20th century, whose average heights surpassed 182.5 cm, and the shortest were women born in Guatemala in 1896 (140.3 cm; 135.8– 144.8). The height differential between the tallest and shortest populations was 19-20 cm a century ago, and has remained the same for women and increased for men a century later despite substantial changes in the ranking of countries.
|
|
| 7. |
- Brant, Luisa C. C., et al.
(author)
-
Association Between Electrocardiographic Age and Cardiovascular Events in Community Settings : The Framingham Heart Study
- 2023
-
In: Circulation. Cardiovascular Quality and Outcomes. - : Ovid Technologies (Wolters Kluwer Health). - 1941-7713 .- 1941-7705. ; 16:7
-
Journal article (peer-reviewed)abstract
- BACKGROUND: Deep neural networks have been used to estimate age from ECGs, the electrocardiographic age (ECG-age), which predicts adverse outcomes. However, this prediction ability has been restricted to clinical settings or relatively short periods. We hypothesized that ECG-age is associated with death and cardiovascular outcomes in the long-standing community-based FHS (Framingham Heart Study).METHODS: We tested the association of ECG-age with chronological age in the FHS cohorts in ECGs from 1986 to 2021. We calculated the gap between chronological and ECG-age (& UDelta;age) and classified individuals as having normal, accelerated, or decelerated aging, if & UDelta;age was within, higher, or lower than the mean absolute error of the model, respectively. We assessed the associations of & UDelta;age, accelerated and decelerated aging with death or cardiovascular outcomes (atrial fibrillation, myocardial infarction, and heart failure) using Cox proportional hazards models adjusted for age, sex, and clinical factors.RESULTS:The study population included 9877 FHS participants (mean age, 55 & PLUSMN;13 years; 54.9% women) with 34 948 ECGs. ECG-age was correlated to chronological age (r=0.81; mean absolute error, 9 & PLUSMN;7 years). After 17 & PLUSMN;8 years of follow-up, every 10-year increase of & UDelta;age was associated with 18% increase in all-cause mortality (hazard ratio [HR], 1.18 [95% CI, 1.12-1.23]), 23% increase in atrial fibrillation risk (HR, 1.23 [95% CI, 1.17-1.29]), 14% increase in myocardial infarction risk (HR, 1.14 [95% CI, 1.05-1.23]), and 40% increase in heart failure risk (HR, 1.40 [95% CI, 1.30-1.52]), in multivariable models. In addition, accelerated aging was associated with a 28% increase in all-cause mortality (HR, 1.28 [95% CI, 1.14-1.45]), whereas decelerated aging was associated with a 16% decrease (HR, 0.84 [95% CI, 0.74-0.95]).CONCLUSIONS:ECG-age was highly correlated with chronological age in FHS. The difference between ECG-age and chronological age was associated with death, myocardial infarction, atrial fibrillation, and heart failure. Given the wide availability and low cost of ECG, ECG-age could be a scalable biomarker of cardiovascular risk.
|
|
| 8. |
|
|
| 9. |
- Habineza, Theogene, et al.
(author)
-
End-to-end risk prediction of atrial fibrillation from the 12-Lead ECG by deep neural networks
- 2023
-
In: Journal of Electrocardiology. - : Elsevier. - 0022-0736 .- 1532-8430. ; 81, s. 193-200
-
Journal article (peer-reviewed)abstract
- Background: Atrial fibrillation (AF) is one of the most common cardiac arrhythmias that affects millions of people each year worldwide and it is closely linked to increased risk of cardiovas-cular diseases such as stroke and heart failure. Machine learning methods have shown promising results in evaluating the risk of developing atrial fibrillation from the electrocardiogram. We aim to develop and evaluate one such algorithm on a large CODE dataset collected in Brazil.Methods: We used the CODE cohort to develop and test a model for AF risk prediction for individual patients from the raw ECG recordings without the use of additional digital biomarkers. The cohort is a collection of ECG recordings and annotations by the Telehealth Network of Minas Gerais, in Brazil. A convolutional neural network based on a residual network architecture was implemented to produce class probabilities for the classification of AF. The probabilities were used to develop a Cox proportional hazards model and a Kaplan-Meier model to carry out survival analysis. Hence, our model is able to perform risk prediction for the development of AF in patients without the condition.Results: The deep neural network model identified patients without indication of AF in the presented ECG but who will develop AF in the future with an AUC score of 0.845. From our survival model, we obtain that patients in the high-risk group (i.e. with the probability of a future AF case being >0.7) are 50% more likely to develop AF within 40 weeks, while patients belonging to the minimal-risk group (i.e. with the probability of a future AF case being less than or equal to 0.1) have >85% chance of remaining AF free up until after seven years.Conclusion: We developed and validated a model for AF risk prediction. If applied in clinical practice, the model possesses the potential of providing valuable and useful information in decision- making and patient management processes.
|
|
| 10. |
- Jidling, Carl, et al.
(author)
-
Screening for Chagas disease from the electrocardiogram using a deep neural network
- 2023
-
In: PLoS Neglected Tropical Diseases. - : Public Library of Science (PLoS). - 1935-2727 .- 1935-2735. ; 17:7
-
Journal article (peer-reviewed)abstract
- Chagas disease (ChD) is a neglected tropical disease, and the diagnosis relies on blood testing of patients from endemic areas. However, there is no clear recommendation on how to select patients for testing in endemic regions. Since most cases of Chronic ChD are asymptomatic, the diagnostic rates are low, preventing patients from receiving adequate treatment.The Electrocardiogram (ECG) is a widely available, low-cost exam, often available in primary care settings. We present an Artificial intelligence (AI) model for automatically detecting ChD from the ECG. AI algorithms have allowed the detection of hidden conditions on the ECG and, to the best of our knowledge, this is the first study that does it for ChD. We utilize large cohorts of patients from the relevant population of all-comers in affected regions in Brazil to develop a model for ChD detection that is then validated on datasets with ground truth labels obtained directly from the patients’ serological status.Our findings demonstrate a promising AI-ECG-based model for discriminating patients with chronic Chagas cardiomyopathy (CCC). The capacity of detecting ChD patients without CCC is still limited. But we believe this can be improved with the addition of epidemiological questions, and that such models can become useful tools for pre-selecting patients for further testing.
|
|
| 11. |
- Lindow, Thomas, et al.
(author)
-
Heart age gap estimated by explainable advanced electrocardiography is associated with cardiovascular risk factors and survival
- 2023
-
In: The European Heart Journal - Digital Health. - : Oxford University Press. - 2634-3916. ; 4:5, s. 384-392
-
Journal article (peer-reviewed)abstract
- AimsDeep neural network artificial intelligence (DNN-AI)-based Heart Age estimations have been presented and used to show that the difference between an electrocardiogram (ECG)-estimated Heart Age and chronological age is associated with prognosis. An accurate ECG Heart Age, without DNNs, has been developed using explainable advanced ECG (A-ECG) methods. We aimed to evaluate the prognostic value of the explainable A-ECG Heart Age and compare its performance to a DNN-AI Heart Age.Methods and resultsBoth A-ECG and DNN-AI Heart Age were applied to patients who had undergone clinical cardiovascular magnetic resonance imaging. The association between A-ECG or DNN-AI Heart Age Gap and cardiovascular risk factors was evaluated using logistic regression. The association between Heart Age Gaps and death or heart failure (HF) hospitalization was evaluated using Cox regression adjusted for clinical covariates/comorbidities. Among patients [n = 731, 103 (14.1%) deaths, 52 (7.1%) HF hospitalizations, median (interquartile range) follow-up 5.7 (4.7-6.7) years], A-ECG Heart Age Gap was associated with risk factors and outcomes [unadjusted hazard ratio (HR) (95% confidence interval) (5 year increments): 1.23 (1.13-1.34) and adjusted HR 1.11 (1.01-1.22)]. DNN-AI Heart Age Gap was associated with risk factors and outcomes after adjustments [HR (5 year increments): 1.11 (1.01-1.21)], but not in unadjusted analyses [HR 1.00 (0.93-1.08)], making it less easily applicable in clinical practice.ConclusionA-ECG Heart Age Gap is associated with cardiovascular risk factors and HF hospitalization or death. Explainable A-ECG Heart Age Gap has the potential for improving clinical adoption and prognostic performance compared with existing DNN-AI-type methods. Graphical Abstract
|
|
| 12. |
- Pastika, Libor, et al.
(author)
-
Artificial intelligence-enhanced electrocardiography derived body mass index as a predictor of future cardiometabolic disease
- 2024
-
In: npj Digital Medicine. - : Springer Nature. - 2398-6352. ; 7
-
Journal article (peer-reviewed)abstract
- The electrocardiogram (ECG) can capture obesity-related cardiac changes. Artificial intelligence-enhanced ECG (AI-ECG) can identify subclinical disease. We trained an AI-ECG model to predict body mass index (BMI) from the ECG alone. Developed from 512,950 12-lead ECGs from the Beth Israel Deaconess Medical Center (BIDMC), a secondary care cohort, and validated on UK Biobank (UKB) (n = 42,386), the model achieved a Pearson correlation coefficient (r) of 0.65 and 0.62, and an R2 of 0.43 and 0.39 in the BIDMC cohort and UK Biobank, respectively for AI-ECG BMI vs. measured BMI. We found delta-BMI, the difference between measured BMI and AI-ECG-predicted BMI (AI-ECG-BMI), to be a biomarker of cardiometabolic health. The top tertile of delta-BMI showed increased risk of future cardiometabolic disease (BIDMC: HR 1.15, p < 0.001; UKB: HR 1.58, p < 0.001) and diabetes mellitus (BIDMC: HR 1.25, p < 0.001; UKB: HR 2.28, p < 0.001) after adjusting for covariates including measured BMI. Significant enhancements in model fit, reclassification and improvements in discriminatory power were observed with the inclusion of delta-BMI in both cohorts. Phenotypic profiling highlighted associations between delta-BMI and cardiometabolic diseases, anthropometric measures of truncal obesity, and pericardial fat mass. Metabolic and proteomic profiling associates delta-BMI positively with valine, lipids in small HDL, syntaxin-3, and carnosine dipeptidase 1, and inversely with glutamine, glycine, colipase, and adiponectin. A genome-wide association study revealed associations with regulators of cardiovascular/metabolic traits, including SCN10A, SCN5A, EXOG and RXRG. In summary, our AI-ECG-BMI model accurately predicts BMI and introduces delta-BMI as a non-invasive biomarker for cardiometabolic risk stratification.
|
|
| 13. |
- Sangha, Veer, et al.
(author)
-
Automated multilabel diagnosis on electrocardiographic images and signals
- 2022
-
In: Nature Communications. - : Springer Nature. - 2041-1723. ; 13:1
-
Journal article (peer-reviewed)abstract
- The application of artificial intelligence for automated diagnosis of electrocardiograms can improve care in remote settings but is limited by the reliance on infrequently available signal-based data. Here, the authors report the development of a multi-label automated diagnosis model for electrocardiographic images. The application of artificial intelligence (AI) for automated diagnosis of electrocardiograms (ECGs) can improve care in remote settings but is limited by the reliance on infrequently available signal-based data. We report the development of a multilabel automated diagnosis model for electrocardiographic images, more suitable for broader use. A total of 2,228,236 12-lead ECGs signals from 811 municipalities in Brazil are transformed to ECG images in varying lead conformations to train a convolutional neural network (CNN) identifying 6 physician-defined clinical labels spanning rhythm and conduction disorders, and a hidden label for gender. The image-based model performs well on a distinct test set validated by at least two cardiologists (average AUROC 0.99, AUPRC 0.86), an external validation set of 21,785 ECGs from Germany (average AUROC 0.97, AUPRC 0.73), and printed ECGs, with performance superior to signal-based models, and learning clinically relevant cues based on Grad-CAM. The model allows the application of AI to ECGs across broad settings.
|
|
| 14. |
- Sangha, Veer, et al.
(author)
-
Detection of Left Ventricular Systolic Dysfunction From Electrocardiographic Images
- 2023
-
In: Circulation. - : Wolters Kluwer. - 0009-7322 .- 1524-4539. ; 148:9, s. 765-777
-
Journal article (peer-reviewed)abstract
- BACKGROUND: Left ventricular (LV) systolic dysfunction is associated with a >8-fold increased risk of heart failure and a 2-fold risk of premature death. The use of ECG signals in screening for LV systolic dysfunction is limited by their availability to clinicians. We developed a novel deep learning-based approach that can use ECG images for the screening of LV systolic dysfunction.METHODS: Using 12-lead ECGs plotted in multiple different formats, and corresponding echocardiographic data recorded within 15 days from the Yale New Haven Hospital between 2015 and 2021, we developed a convolutional neural network algorithm to detect an LV ejection fraction <40%. The model was validated within clinical settings at Yale New Haven Hospital and externally on ECG images from Cedars Sinai Medical Center in Los Angeles, CA; Lake Regional Hospital in Osage Beach, MO; Memorial Hermann Southeast Hospital in Houston, TX; and Methodist Cardiology Clinic of San Antonio, TX. In addition, it was validated in the prospective Brazilian Longitudinal Study of Adult Health. Gradient-weighted class activation mapping was used to localize class-discriminating signals on ECG images.RESULTS: Overall, 385 601 ECGs with paired echocardiograms were used for model development. The model demonstrated high discrimination across various ECG image formats and calibrations in internal validation (area under receiving operation characteristics [AUROCs], 0.91; area under precision-recall curve [AUPRC], 0.55); and external sets of ECG images from Cedars Sinai (AUROC, 0.90 and AUPRC, 0.53), outpatient Yale New Haven Hospital clinics (AUROC, 0.94 and AUPRC, 0.77), Lake Regional Hospital (AUROC, 0.90 and AUPRC, 0.88), Memorial Hermann Southeast Hospital (AUROC, 0.91 and AUPRC 0.88), Methodist Cardiology Clinic (AUROC, 0.90 and AUPRC, 0.74), and Brazilian Longitudinal Study of Adult Health cohort (AUROC, 0.95 and AUPRC, 0.45). An ECG suggestive of LV systolic dysfunction portended >27-fold higher odds of LV systolic dysfunction on transthoracic echocardiogram (odds ratio, 27.5 [95% CI, 22.3-33.9] in the held-out set). Class-discriminative patterns localized to the anterior and anteroseptal leads (V2 and V3), corresponding to the left ventricle regardless of the ECG layout. A positive ECG screen in individuals with an LV ejection fraction & GE;40% at the time of initial assessment was associated with a 3.9-fold increased risk of developing incident LV systolic dysfunction in the future (hazard ratio, 3.9 [95% CI, 3.3-4.7]; median follow-up, 3.2 years).CONCLUSIONS: We developed and externally validated a deep learning model that identifies LV systolic dysfunction from ECG images. This approach represents an automated and accessible screening strategy for LV systolic dysfunction, particularly in low-resource settings.
|
|
| 15. |
- Zhang, Cuili, et al.
(author)
-
Association of lifestyle with deep learning predicted electrocardiographic age
- 2023
-
In: Frontiers in Cardiovascular Medicine. - : Frontiers Media S.A.. - 2297-055X. ; 10
-
Journal article (peer-reviewed)abstract
- Background: People age at different rates. Biological age is a risk factor for many chronic diseases independent of chronological age. A good lifestyle is known to improve overall health, but its association with biological age is unclear.Methods: This study included participants from the UK Biobank who had undergone 12-lead resting electrocardiography (ECG). Biological age was estimated by a deep learning model (defined as ECG-age), and the difference between ECG-age and chronological age was defined as Delta age. Participants were further categorized into an ideal (score 4), intermediate (scores 2 and 3) or unfavorable lifestyle (score 0 or 1). Four lifestyle factors were investigated, including diet, alcohol consumption, physical activity, and smoking. Linear regression models were used to examine the association between lifestyle factors and Delta age, and the models were adjusted for sex and chronological age.Results: This study included 44,094 individuals (mean age 64 +/- 8, 51.4% females). A significant correlation was observed between predicted biological age and chronological age (correlation coefficient = 0.54, P < 0.001) and the mean Delta age (absolute error of biological age and chronological age) was 9.8 +/- 7.4 years. Delta age was significantly associated with all of the four lifestyle factors, with the effect size ranging from 0.41 +/- 0.11 for the healthy diet to 2.37 +/- 0.30 for non-smoking. Compared with an ideal lifestyle, an unfavorable lifestyle was associated with an average of 2.50 +/- 0.29 years of older predicted ECG-age.Conclusion: In this large contemporary population, a strong association was observed between all four studied healthy lifestyle factors and deaccelerated aging. Our study underscores the importance of a healthy lifestyle to reduce the burden of aging-related diseases.
|
|
| 16. |
- Zvuloni, Eran, et al.
(author)
-
On Merging Feature Engineering and Deep Learning for Diagnosis, Risk Prediction and Age Estimation Based on the 12-Lead ECG
- 2023
-
In: IEEE Transactions on Biomedical Engineering. - : Institute of Electrical and Electronics Engineers (IEEE). - 0018-9294 .- 1558-2531. ; 70:7, s. 2227-2236
-
Journal article (peer-reviewed)abstract
- Objective: Over the past few years, deep learning (DL) has been used extensively in research for 12-lead electrocardiogram (ECG) analysis. However, it is unclear whether the explicit or implicit claims made on DL superiority to the more classical feature engineering (FE) approaches, based on domain knowledge, hold. In addition, it remains unclear whether combining DL with FE may improve performance over a single modality.Methods: To address these research gaps and in-line with recent major experiments, we revisited three tasks: cardiac arrhythmia diagnosis (multiclass-multilabel classification), atrial fibrillation risk prediction (binary classification), and age estimation (regression). We used an overall dataset of 2.3M 12-lead ECG recordings to train the following models for each task: i) a random forest taking FE as input; ii) an end-to-end DL model; and iii) a merged model of FE+DL.Results: FE yielded comparable results to DL while necessitating significantly less data for the two classification tasks. DL outperformed FE for the regression task. For all tasks, merging FE with DL did not improve performance over DL alone. These findings were confirmed on the additional PTB-XL dataset. Conclusion: We found that for traditional 12-lead ECG based diagnosis tasks, DL did not yield a meaningful improvement over FE, while it improved significantly the nontraditional regression task. We also found that combining FE with DL did not improve over DL alone, which suggests that the FE was redundant with the features learned by DL.Significance: Our findings provides important recommendations on 12-lead ECG based machine learning strategy and data regime to choose for a given task. When looking at maximizing performance as the end goal, if the task is nontraditional and a large dataset is available then DL is preferable. If the task is a classical one and/or a small dataset is available then a FE approach may be the better choice.
|
|
| 17. |
- Andersson, Carl, et al.
(author)
-
Deep convolutional networks in system identification
- 2019
-
In: Proc. 58th IEEE Conference on Decision and Control. - : IEEE. - 9781728113982 ; , s. 3670-3676
-
Conference paper (peer-reviewed)abstract
- Recent developments within deep learning are relevant for nonlinear system identification problems. In this paper, we establish connections between the deep learning and the system identification communities. It has recently been shown that convolutional architectures are at least as capable as recurrent architectures when it comes to sequence modeling tasks. Inspired by these results we explore the explicit relationships between the recently proposed temporal convolutional network (TCN) and two classic system identification model structures; Volterra series and block-oriented models. We end the paper with an experimental study where we provide results on two real-world problems, the well-known Silverbox dataset and a newer dataset originating from ground vibration experiments on an F-16 fighter aircraft.
|
|
| 18. |
- Gedon, Daniel, 1994-, et al.
(author)
-
First Steps Towards Self-Supervised Pretraining of the 12-Lead ECG
- 2021
-
In: 2021 Computing In Cardiology (CINC). - : Institute of Electrical and Electronics Engineers (IEEE). - 9781665479165
-
Conference paper (peer-reviewed)abstract
- Self-supervised learning is a paradigm that extracts general features which describe the input space by artificially generating labels from the input without the need for explicit annotations. The learned features can then be used by transfer learning to boost the performance on a downstream task. Such methods have recently produced state of the art results in natural language processing and computer vision. Here, we propose a self-supervised learning method for 12-lead electrocardiograms (ECGs). For pretraining the model we design a task to mask out subsegements of all channels of the input signals and try to predict the actual values. As the model architecture, we use a U-ResNet containing an encoder-decoder structure. We test our method by self-supervised pretraining on the CODE dataset and then transfer the learnt features by finetuning on the PTBXL and CPSC benchmarks to evaluate the effect of our method in the classification of 12-leads ECGs. The method does provide modest improvements in performance when compared to not using pretraining. In future work we will make use of these ideas in smaller dataset, where we believe it can lead to larger performance gains.
|
|
| 19. |
- Gedon, Daniel, 1994-, et al.
(author)
-
No Double Descent in Principal Component Regression: A High-Dimensional Analysis
-
Other publication (other academic/artistic)abstract
- Understanding the generalization properties of large-scale models necessitates incorporating realistic data assumptions into the analysis. Therefore, we consider Principal Component Regression (PCR)---combining principal component analysis and linear regression---on data from a low-dimensional manifold. We present an analysis of PCR when the data is sampled from a spiked covariance model, obtaining fundamental asymptotic guarantees for the generalization risk of this model. Our analysis is based on random matrix theory and allows us to provide guarantees for high-dimensional data. We additionally present an analysis of the distribution shift between training and test data. The results allow us to disentangle the effects of (1) the number of parameters, (2) the data-generating model and, (3) model misspecification on the generalization risk. The use of PCR effectively regularizes the model and prevents the interpolation peak of the double descent. Our theoretical findings are empirically validated in simulation, demonstrating their practical relevance.
|
|
| 20. |
- Gedon, Daniel, 1994-
(author)
-
On Deep Learning for Low-Dimensional Representations
- 2024
-
Doctoral thesis (other academic/artistic)abstract
- In science and engineering, we are often concerned with creating mathematical models from data. These models are abstractions of observed real-world processes where the goal is often to understand these processes or to use the models to predict future instances of the observed process. Natural processes often exhibit low-dimensional structures which we can embed into the model. In mechanistic models, we directly include this structure into the model through mathematical equations often inspired by physical constraints. In contrast, within machine learning and particularly in deep learning we often deal with high-dimensional data such as images and learn a model without imposing a low-dimensional structure. Instead, we learn some kind of representations that are useful for the task at hand. While representation learning arguably enables the power of deep neural networks, it is less clear how to understand real-world processes from these models or whether we can benefit from including a low-dimensional structure in the model.Learning from data with intrinsic low-dimensional structure and how to replicate this structure in machine learning models is studied within this dissertation. While we put specific emphasis on deep neural networks, we also consider kernel machines in the context of Gaussian processes, as well as linear models, for example by studying the generalisation of models with an explicit low-dimensional structure. First, we argue that many real-world observations have an intrinsic low-dimensional structure. We can find evidence of this structure for example through low-rank approximations of many real-world data sets. Then, we face two open-ended research questions. First, we study the behaviour of machine learning models when they are trained on data with low-dimensional structures. Here we investigate fundamental aspects of learning low-dimensional representations and how well models with explicit low-dimensional structures perform. Second, we focus on applications in the modelling of dynamical systems and the medical domain. We investigate how we can benefit from low-dimensional representations for these applications and explore the potential of low-dimensional model structures for predictive tasks. Finally, we give a brief outlook on how we go beyond learning low-dimensional structures and identify the underlying mechanisms that generate the data to better model and understand these processes.This dissertation provides an overview of learning low-dimensional structures in machine learning models. It covers a wide range of topics from representation learning over the study of generalisation in overparameterized models to applications with time series and medical applications. However, each contribution opens up a range of questions to study in the future. Therefore this dissertation serves as a starting point to further explore learning of low-dimensional structure and representations.
|
|
| 21. |
- Gustafsson, Stefan, et al.
(author)
-
Development and validation of deep learning ECG-based prediction of myocardial infarction in emergency department patients
- 2022
-
In: Scientific Reports. - : Springer Nature. - 2045-2322. ; 12
-
Journal article (peer-reviewed)abstract
- Myocardial infarction diagnosis is a common challenge in the emergency department. In managed settings, deep learning-based models and especially convolutional deep models have shown promise in electrocardiogram (ECG) classification, but there is a lack of high-performing models for the diagnosis of myocardial infarction in real-world scenarios. We aimed to train and validate a deep learning model using ECGs to predict myocardial infarction in real-world emergency department patients. We studied emergency department patients in the Stockholm region between 2007 and 2016 that had an ECG obtained because of their presenting complaint. We developed a deep neural network based on convolutional layers similar to a residual network. Inputs to the model were ECG tracing, age, and sex; and outputs were the probabilities of three mutually exclusive classes: non-ST-elevation myocardial infarction (NSTEMI), ST-elevation myocardial infarction (STEMI), and control status, as registered in the SWEDEHEART and other registries. We used an ensemble of five models. Among 492,226 ECGs in 214,250 patients, 5,416 were recorded with an NSTEMI, 1,818 a STEMI, and 485,207 without a myocardial infarction. In a random test set, our model could discriminate STEMIs/NSTEMIs from controls with a C-statistic of 0.991/0.832 and had a Brier score of 0.001/0.008. The model obtained a similar performance in a temporally separated test set of the study sample, and achieved a C-statistic of 0.985 and a Brier score of 0.002 in discriminating STEMIs from controls in an external test set. We developed and validated a deep learning model with excellent performance in discriminating between control, STEMI, and NSTEMI on the presenting ECG of a real-world sample of the important population of all-comers to the emergency department. Hence, deep learning models for ECG decision support could be valuable in the emergency department.
|
|
| 22. |
- Horta Ribeiro, Antônio, et al.
(author)
-
How convolutional neural networks deal with aliasing
- 2021
-
In: 2021 IEEE International Conference On Acoustics, Speech And Signal Processing (ICASSP 2021). - : Institute of Electrical and Electronics Engineers (IEEE). - 9781728176055 ; , s. 2755-2759
-
Conference paper (peer-reviewed)abstract
- The convolutional neural network (CNN) remains an essential tool in solving computer vision problems. Standard convolutional architectures consist of stacked layers of operations that progressively downscale the image. Aliasing is a well-known side-effect of downsampling that may take place: it causes high-frequency components of the original signal to become indistinguishable from its low-frequency components. While downsampling takes place in the max-pooling layers or in the strided-convolutions in these models, there is no explicit mechanism that prevents aliasing from taking place in these layers. Due to the impressive performance of these models, it is natural to suspect that they, somehow, implicitly deal with this distortion. The question we aim to answer in this paper is simply: "how and to what extent do CNNs counteract aliasing?" We explore the question by means of two examples: In the first, we assess the CNNs capability of distinguishing oscillations at the input, showing that the redundancies in the intermediate channels play an important role in succeeding at the task; In the second, we show that an image classifier CNN while, in principle, capable of implementing anti-aliasing filters, does not prevent aliasing from taking place in the intermediate layers.
|
|
| 23. |
- Horta Ribeiro, Antônio, et al.
(author)
-
On the smoothness of nonlinear system identification
- 2020
-
In: Automatica. - : Elsevier BV. - 0005-1098 .- 1873-2836. ; 121
-
Journal article (peer-reviewed)abstract
- We shed new light on the smoothness of optimization problems arising in prediction error parameter estimation of linear and nonlinear systems. We show that for regions of the parameter space where the model is not contractive, the Lipschitz constant and β-smoothness of the objective function might blow up exponentially with the simulation length, making it hard to numerically find minima within those regions or, even, to escape from them. In addition to providing theoretical understanding of this problem, this paper also proposes the use of multiple shooting as a viable solution. The proposed method minimizes the error between a prediction model and the observed values. Rather than running the prediction model over the entire dataset, multiple shooting splits the data into smaller subsets and runs the prediction model over each subset, making the simulation length a design parameter and making it possible to solve problems that would be infeasible using a standard approach. The equivalence to the original problem is obtained by including constraints in the optimization. The new method is illustrated by estimating the parameters of nonlinear systems with chaotic or unstable behavior, as well as neural networks. We also present a comparative analysis of the proposed method with multi-step-ahead prediction error minimization.
|
|
| 24. |
- Horta Ribeiro, Antônio, et al.
(author)
-
Overparameterized Linear Regression Under Adversarial Attacks
- 2023
-
In: IEEE Transactions on Signal Processing. - : Institute of Electrical and Electronics Engineers (IEEE). - 1053-587X .- 1941-0476. ; 71, s. 601-614
-
Journal article (peer-reviewed)abstract
- We study the error of linear regression in the face of adversarial attacks. In this framework, an adversary changes the input to the regression model in order to maximize the prediction error. We provide bounds on the prediction error in the presence of an adversary as a function of the parameter norm and the error in the absence of such an adversary. We show how these bounds make it possible to study the adversarial error using analysis from non-adversarial setups. The obtained results shed light on the robustness of overparameterized linear models to adversarial attacks. Adding features might be either a source of additional robustness or brittleness. On the one hand, we use asymptotic results to illustrate how double-descent curves can be obtained for the adversarial error. On the other hand, we derive conditions under which the adversarial error can grow to infinity as more features are added, while at the same time, the test error goes to zero. We show this behavior is caused by the fact that the norm of the parameter vector grows with the number of features. It is also established that l(infinity) and l(2)-adversarial attacks might behave fundamentally differently due to how the l(1) and l(2)-norms of random projections concentrate. We also show how our reformulation allows for solving adversarial training as a convex optimization problem. This fact is then exploited to establish similarities between adversarial training and parameter-shrinking methods and to study how the training might affect the robustness of the estimated models.
|
|
| 25. |
|
|
