Novel Bloodless Potassium Determination Using a Signal-Processed Single-Lead ECG

A Scoping Review of Varying Mobile Electrocardiographic Devices

The electrocardiogram (ECG) can now be measured using mobile devices. Mobile ECG devices, which are defined as devices capable of recording and transmitting non-standard ECGs, offer numerous advantages such as cost-effectiveness and being user-friendly. Mobile ECG can also extend recording lengths (e.g., 2 days, 14 days), which is necessary to capture important intermittent events (e.g., cardiac arrhythmias) and evaluate prognostic risk markers (e.g., prolonged corrected QT (QTc) interval). Some mobile ECG devices can even connect to broadband networks allowing patients to remotely transmit their ECG to a clinician. This article systematically examines different mobile ECG devices used in prior studies and provides a detailed assessment of five diverse yet commonly used mobile ECG devices: AliveCor KardiaMobile; AliveCor KardiaMobile 6L; iRhythm ZioPatch; Apple Smartwatch ECG; and CardioSecur System. These mobile ECG devices are diverse in the number of leads measured and the duration of monitoring. Similar to their diversity, there has been a wide range of clinical applications of mobile ECG devices. Despite significant progress, questions regarding data quality, and clinican and patient acceptance and compliance persist.

An Artificial Intelligence Analysis of Electrocardiograms for the Clinical Diagnosis of Cardiovascular Diseases: A Narrative Review

Artificial intelligence (AI) applied to cardiovascular disease (CVD) is enjoying great success in the field of scientific research. Electrocardiograms (ECGs) are the cornerstone form of examination in cardiology and are the most widely used diagnostic tool because they are widely available, inexpensive, and fast. Applications of AI to ECGs, especially deep learning (DL) methods using convolutional neural networks (CNNs), have been developed in many fields of cardiology in recent years. Deep learning methods provide valuable support for rapid ECG interpretation, demonstrating a diagnostic capability overlapping with specialists in the diagnosis of CVD by a classical analysis of macroscopic changes in the ECG trace. Through photoplethysmography, wearable devices can obtain single-derivative ECGs for the recognition of AI-diagnosed arrhythmias. In addition, CNNs have been developed that recognize no macroscopic electrocardiographic changes and can predict, from a 12-lead ECG, atrial fibrillation, even from sinus rhythm; left and right ventricular function; hypertrophic cardiomyopathy; acute coronary syndromes; or aortic stenosis. The fields of application are many, but numerous are the limitations, mainly associated with the reliability of the acquired data, an inability to verify black box processes, and medico-legal and ethical problems. The challenge of modern medicine is to recognize the limitations of AI and overcome them.

Hyperkalemia Detection in Emergency Departments Using Initial ECGs: A Smartphone AI ECG Analyzer vs. Board-Certified Physicians

BACKGROUND

Hyperkalemia is a potentially fatal condition that mandates rapid identification in emergency departments (EDs). Although a 12-lead electrocardiogram (ECG) can indicate hyperkalemia, subtle changes in the ECG often pose detection challenges. An artificial intelligence application that accurately assesses hyperkalemia risk from ECGs could revolutionize patient screening and treatment. We aimed to evaluate the efficacy and reliability of a smartphone application, which utilizes camera-captured ECG images, in quantifying hyperkalemia risk compared to human experts.

METHODS

We performed a retrospective analysis of ED hyperkalemic patients (serum potassium ≥ 6 mmol/L) and their age- and sex-matched non-hyperkalemic controls. The application was tested by five users and its performance was compared to five board-certified emergency physicians (EPs).

RESULTS

Our study included 125 patients. The area under the curve (AUC)-receiver operating characteristic of the application's output was nearly identical among the users, ranging from 0.898 to 0.904 (median: 0.902), indicating almost perfect interrater agreement (Fleiss' kappa 0.948). The application demonstrated high sensitivity (0.797), specificity (0.934), negative predictive value (NPV) (0.815), and positive predictive value (PPV) (0.927). In contrast, the EPs showed moderate interrater agreement (Fleiss' kappa 0.551), and their consensus score had a significantly lower AUC of 0.662. The physicians' consensus demonstrated a sensitivity of 0.203, specificity of 0.934, NPV of 0.527, and PPV of 0.765. Notably, this performance difference remained significant regardless of patients' sex and age (P < 0.001 for both).

CONCLUSION

Our findings suggest that a smartphone application can accurately and reliably quantify hyperkalemia risk using initial ECGs in the ED.

Comparison of quantifiable electrocardiographic changes associated with severe hyperkalemia

BACKGROUND

Hyperkalemia (HK) is a life-threatening condition that is frequently evaluated by electrocardiogram (ECG). ECG changes in severe HK (≥ 6.3 mEq/L) are not well-characterized. This study sought to compare and correlate ECG metrics in severe HK to baseline normokalemic ECGs and serum potassium.

METHODS

A retrospective analysis of 340 severe HK encounters with corresponding normokalemic ECGs was performed.

RESULTS

Various ECG metrics were analyzed. P wave amplitude in lead II, QRS duration, T wave slope, ratio of T wave amplitude: duration, and ratios of T wave: QRS amplitudes were significantly different between normokalemic and HK ECGs. P wave amplitude attenuation in lead II correlated better with serum potassium than in V1. T wave metrics that incorporated both T wave and QRS amplitudes correlated better than metrics utilizing T wave metrics alone.

CONCLUSION

Multiple statistically significant and quantifiable differences among ECG metrics were observed between normokalemic and HK ECGs and correlated with increasing degrees of serum potassium and along the continuum of serum potassium. When incorporated into a logistic regression model, the ability to distinguish HK versus normokalemia on ECG improved significantly. These findings could be integrated into an ECG acquisition system that can more accurately identify severe HK.

A framework for integrating artificial intelligence for clinical care with continuous therapeutic monitoring

The complex relationships between continuously monitored health signals and therapeutic regimens can be modelled via machine learning. However, the clinical implementation of the models will require changes to clinical workflows. Here we outline ClinAIOps ('clinical artificial-intelligence operations'), a framework that integrates continuous therapeutic monitoring and the development of artificial intelligence (AI) for clinical care. ClinAIOps leverages three feedback loops to enable the patient to make treatment adjustments using AI outputs, the clinician to oversee patient progress with AI assistance, and the AI developer to receive continuous feedback from both the patient and the clinician. We lay out the central challenges and opportunities in the deployment of ClinAIOps by means of examples of its application in the management of blood pressure, diabetes and Parkinson's disease. By enabling more frequent and accurate measurements of a patient's health and more timely adjustments to their treatment, ClinAIOps may substantially improve patient outcomes.

Differences in ventricular wall composition may explain inter-patient variability in the ECG response to variations in serum potassium and calcium

Objective: Chronic kidney disease patients have a decreased ability to maintain normal electrolyte concentrations in their blood, which increases the risk for ventricular arrhythmias and sudden cardiac death. Non-invasive monitoring of serum potassium and calcium concentration, [K+] and [Ca2+], can help to prevent arrhythmias in these patients. Electrocardiogram (ECG) markers that significantly correlate with [K+] and [Ca2+] have been proposed, but these relations are highly variable between patients. We hypothesized that inter-individual differences in cell type distribution across the ventricular wall can help to explain this variability. Methods: A population of human heart-torso models were built with different proportions of endocardial, midmyocardial and epicardial cells. Propagation of ventricular electrical activity was described by a reaction-diffusion model, with modified Ten Tusscher-Panfilov dynamics. [K+] and [Ca2+] were varied individually and in combination. Twelve-lead ECGs were simulated and the width, amplitude and morphological variability of T waves and QRS complexes were quantified. Results were compared to measurements from 29 end-stage renal disease (ESRD) patients undergoing hemodialysis (HD). Results: Both simulations and patients data showed that most of the analyzed T wave and QRS complex markers correlated strongly with [K+] (absolute median Pearson correlation coefficients, r, ranging from 0.68 to 0.98) and [Ca2+] (ranging from 0.70 to 0.98). The same sign and similar magnitude of median r was observed in the simulations and the patients. Different cell type distributions in the ventricular wall led to variability in ECG markers that was accentuated at high [K+] and low [Ca2+], in agreement with the larger variability between patients measured at the onset of HD. The simulated ECG variability explained part of the measured inter-patient variability. Conclusion: Changes in ECG markers were similarly related to [K+] and [Ca2+] variations in our models and in the ESRD patients. The high inter-patient ECG variability may be explained by variations in cell type distribution across the ventricular wall, with high sensitivity to variations in the proportion of epicardial cells. Significance: Differences in ventricular wall composition help to explain inter-patient variability in ECG response to [K+] and [Ca2+]. This finding can be used to improve serum electrolyte monitoring in ESRD patients.

Using Deep Transfer Learning to Detect Hyperkalemia From Ambulatory Electrocardiogram Monitors in Intensive Care Units: Personalized Medicine Approach

BACKGROUND

Hyperkalemia is a critical condition, especially in intensive care units. So far, there have been no accurate and noninvasive methods for recognizing hyperkalemia events on ambulatory electrocardiogram monitors.

OBJECTIVE

This study aimed to improve the accuracy of hyperkalemia predictions from ambulatory electrocardiogram (ECG) monitors using a personalized transfer learning method; this would be done by training a generic model and refining it with personal data.

METHODS

This retrospective cohort study used open source data from the Waveform Database Matched Subset of the Medical Information Mart From Intensive Care III (MIMIC-III). We included patients with multiple serum potassium test results and matched ECG data from the MIMIC-III database. A 1D convolutional neural network-based deep learning model was first developed to predict hyperkalemia in a generic population. Once the model achieved a state-of-the-art performance, it was used in an active transfer learning process to perform patient-adaptive heartbeat classification tasks.

RESULTS

The results show that by acquiring data from each new patient, the personalized model can improve the accuracy of hyperkalemia detection significantly, from an average of 0.604 (SD 0.211) to 0.980 (SD 0.078), when compared with the generic model. Moreover, the area under the receiver operating characteristic curve level improved from 0.729 (SD 0.240) to 0.945 (SD 0.094).

CONCLUSIONS

By using the deep transfer learning method, we were able to build a clinical standard model for hyperkalemia detection using ambulatory ECG monitors. These findings could potentially be extended to applications that continuously monitor one's ECGs for early alerts of hyperkalemia and help avoid unnecessary blood tests.

Hyperkalemia and Electrocardiogram Manifestations in End-Stage Renal Disease

Hyperkalemia is one of the more common acute life-threatening metabolic emergencies. The aim of our study is to determine the correlation and accuracy of abnormal ECG parameters as a function of serum potassium concentration in the end-stage renal disease (ESRD) population. We performed a retrospective chart review of emergency department patients presenting with ESRD and receiving emergent hemodialysis treatment. A total of 96 patients, each with five independent ED visits, provided 480 sets of ECGs and electrolytes. Of these, four ECGs were excluded for inability to interpret, leaving a total of 476 patient encounters that met all inclusion criteria. Linear regression analysis on the limited data set for serum potassium versus T/R in V2, V3, and V4, PR, and QRS found weak correlations (r2 = 0.02 to 0.12) with statistical significance <0.05 level for T/R in V2, V3, and V4. In summary, we found that a QRS duration of 120 ms or greater is most predictive of hyperkalemia in the ESRD population. On the other hand, T/R ratio, PR interval and QRS duration have poor correlations with serum potassium and are not predictive of hyperkalemia in patients with ESRD.

Development and validation of a dynamic deep learning algorithm using electrocardiogram to predict dyskalaemias in patients with multiple visits

Aims

Deep learning models (DLMs) have shown superiority in electrocardiogram (ECG) analysis and have been applied to diagnose dyskalaemias. However, no study has explored the performance of DLM-enabled ECG in continuous follow-up scenarios. Therefore, we proposed a dynamic revision of DLM-enabled ECG to use personal pre-annotated ECGs to enhance the accuracy in patients with multiple visits.

Methods and results

We retrospectively collected 168 450 ECGs with corresponding serum potassium (K⁺) levels from 103 091 patients as development samples. In the internal/external validation sets, the numbers of ECGs with corresponding K⁺ were 37 246/47 604 from 13 555/20 058 patients. Our dynamic revision method showed better performance than the traditional direct prediction for diagnosing hypokalaemia [area under the receiver operating characteristic curve (AUC) = 0.730/0.720-0.788/0.778] and hyperkalaemia (AUC = 0.884/0.888-0.915/0.908) in patients with multiple visits.

Conclusion

Our method has shown a distinguishable improvement in DLMs for diagnosing dyskalaemias in patients with multiple visits, and we also proved its application in ejection fraction prediction, which could further improve daily clinical practice.

Initial and terminal T wave angle as hyperkalemia indicator in chronic kidney disease

BACKGROUND

Hyperkalemia is one prevalent complication in chronic kidney disease and is considered fatal since it potentially causes malignant arrhythmias and mortality. It is associated with electrocardiography (ECG) changes, such as peaked T wave in all ECG leads. However, the universal definition of the peaked T wave is still unclear, with low sensitivity and specificity.

AIM

: This study aims to determine the predictive value of initial and terminal T wave angle in detecting hyperkalemia among CKD patients.

METHODS

: A cross-sectional study was conducted at Dr. Soetomo General Hospital, including all adult hospitalized CKD patients. A caliper was used to measure T wave morphology. The initial deflection angle (Tia) and terminal deflection angle (Tta) were calculated from an arctan of T peak amplitude and the respective initial or terminal length. The receiver operating characteristics (ROC) curve was analyzed to determine the area under the curve (AUC) and optimal cut-off.

RESULTS

: A total of 220 CKD patients were enrolled in this study, with 98 patients with hyperkalemia (potassium >5.0). The majority of the patients were male, with a mean age of 51.12±12.58 years. Ti-Tp duration, Tp-Tt duration, Tia, Tta, and Tp amplitude were significantly higher in the hyperkalemia group (all p<0.05). A Spearman correlation analysis demonstrated a significant positive correlation of Tia (r=0.346 and p<0.001) and Tta (r=0.445 and p<0.001) with potassium levels in the participants. The optimal cut-off angle for Tta was 66.20^o (sensitivity=67.3% and specificity=73.8%) and Tia was 61.07^o (sensitivity=66.3% and specificity=69.7%).

CONCLUSION

: The terminal T wave angle outperformed the initial angle in predicting hyperkalemia in CKD patients.

Artificial Intelligence-Enabled ECG: Physiologic and Pathophysiologic Insights and Implications

Advancements in machine learning and computing methods have given new life and great excitement to one of the most essential diagnostic tools to date-the electrocardiogram (ECG). The application of artificial intelligence-enabled ECG (AI-ECG) has resulted in the ability to identify electrocardiographic signatures of conventional and unique variables and pathologies, giving way to tremendous clinical potential. However, what these AI-ECG models are detecting that the human eye is missing remains unclear. In this article, we highlight some of the recent developments in the field and their potential clinical implications, while also attempting to shed light on the physiologic and pathophysiologic features that enable these models to have such high diagnostic yield. © 2022 American Physiological Society. Compr Physiol 12:3417-3424, 2022.

[Artificial intelligence-based ECG analysis: current status and future perspectives : Part 2: Recent studies and future]

Während grundlegende Aspekte der Anwendung von künstlicher Intelligenz (KI) zur Elektrokardiogramm(EKG)-Analyse in Teil 1 dieser Übersicht behandelt wurden, beschäftigt sich die vorliegende Arbeit (Teil 2) mit einer Besprechung von aktuellen Studien zum praktischen Einsatz dieser neuen Technologien und Aspekte ihrer aktuellen und möglichen zukünftigen Anwendung. Die Anzahl der zum Thema KI-basierte EKG-Analyse publizierten Studien steigt seit 2017 rasant an. Dies gilt vor allem für Untersuchungen, die Deep Learning (DL) mit künstlichen neuronalen Netzen (KNN) einsetzen. Inhaltlich geht es nicht nur darum, die Schwächen der klassischen EKG-Diagnostik mit Hilfe von KI zu überwinden und die diagnostische Güte des Verfahrens zu verbessern, sondern auch die Funktionalität des EKGs zu erweitern. Angestrebt wird die Erkennung spezieller kardiologischer und nichtkardiologischer Krankheitsbilder sowie die Vorhersage zukünftiger Krankheitszustände, z. B. die zukünftige Entwicklung einer linksventrikulären Dysfunktion oder das zukünftige Auftreten von Vorhofflimmern. Möglich wird dies, indem KI mittels DL in riesigen EKG-Datensätzen subklinische Muster findet und für die Algorithmen-Entwicklung nutzt. Die KI-unterstützte EKG-Analyse wird somit zu einem Screening-Instrument und geht weit darüber hinaus, nur besser als ein Kardiologe zu sein. Die erzielten Fortschritte sind bemerkenswert und sorgen in Fachwelt und Öffentlichkeit für Aufmerksamkeit und Euphorie. Bei den meisten Studien handelt es sich allerdings um Proof-of-Concept-Studien. Häufig werden private (institutionseigene) Daten verwendet, deren Qualität unklar ist. Bislang ist nur selten eine klinische Validierung der entwickelten Algorithmen in anderen Kollektiven und Szenarien erfolgt. Besonders problematisch ist, dass der Weg, wie KI eine Lösung findet, bislang meistens verborgen bleibt (Blackbox-Charakter). Damit steckt die KI-basierte Elektrokardiographie noch in den Kinderschuhen. Unbestritten ist aber schon absehbar, dass das EKG als einfach anzuwendendes und beliebig oft wiederholbares diagnostisches Verfahren auch in Zukunft nicht nur weiterhin unverzichtbar sein wird, sondern durch KI an klinischer Bedeutung gewinnen wird.

Monitoring of Serum Potassium and Calcium Levels in End-Stage Renal Disease Patients by ECG Depolarization Morphology Analysis

Objective: Non-invasive estimation of serum potassium, [K+], and calcium, [Ca2+], can help to prevent life-threatening ventricular arrhythmias in patients with advanced renal disease, but current methods for estimation of electrolyte levels have limitations. We aimed to develop new markers based on the morphology of the QRS complex of the electrocardiogram (ECG). Methods: ECG recordings from 29 patients undergoing hemodialysis (HD) were processed. Mean warped QRS complexes were computed in two-minute windows at the start of an HD session, at the end of each HD hour and 48 h after it. We quantified QRS width, amplitude and the proposed QRS morphology-based markers that were computed by warping techniques. Reference [K+] and [Ca2+] were determined from blood samples acquired at the time points where the markers were estimated. Linear regression models were used to estimate electrolyte levels from the QRS markers individually and in combination with T wave morphology markers. Leave-one-out cross-validation was used to assess the performance of the estimators. Results: All markers, except for QRS width, strongly correlated with [K+] (median Pearson correlation coefficients, r, ranging from 0.81 to 0.87) and with [Ca2+] (r ranging from 0.61 to 0.76). QRS morphology markers showed very low sensitivity to heart rate (HR). Actual and estimated serum electrolyte levels differed, on average, by less than 0.035 mM (relative error of 0.018) for [K+] and 0.010 mM (relative error of 0.004) for [Ca2+] when patient-specific multivariable estimators combining QRS and T wave markers were used. Conclusion: QRS morphological markers allow non-invasive estimation of [K+] and [Ca2+] with low sensitivity to HR. The estimation performance is improved when multivariable models, including T wave markers, are considered. Significance: Markers based on the QRS complex of the ECG could contribute to non-invasive monitoring of serum electrolyte levels and arrhythmia risk prediction in patients with renal disease.

Estimation of potassium levels in hemodialysis patients by T wave nonlinear dynamics and morphology markers

OBJECTIVE

Noninvasive screening of hypo- and hyperkalemia can prevent fatal arrhythmia in end-stage renal disease (ESRD) patients, but current methods for monitoring of serum potassium (K+) have important limitations. We investigated changes in nonlinear dynamics and morphology of the T wave in the electrocardiogram (ECG) of ESRD patients during hemodialysis (HD), assessing their relationship with K+ and designing a K+ estimator.

METHODS

ECG recordings from twenty-nine ESRD patients undergoing HD were processed. T waves in 2-min windows were extracted at each hour during an HD session as well as at 48 h after HD start. T wave nonlinear dynamics were characterized by two indices related to the maximum Lyapunov exponent (λt, λwt) and a divergence-related index (η). Morphological variability in the T wave was evaluated by three time warping-based indices (dw, reflecting morphological variability in the time domain, and da and daNL, in the amplitude domain). K+was measured from blood samples extracted during and after HD. Stage-specific and patient-specific K+ estimators were built based on the quantified indices and leave-one-out cross-validation was performed separately for each of the estimators.

RESULTS

The analyzed indices showed high inter-individual variability in their relationship with K+. Nevertheless, all of them had higher values at the HD start and 48 h after it, corresponding to the highest K+. The indices η and dw were the most strongly correlated with K+ (median Pearson correlation coefficient of 0.78 and 0.83, respectively) and were used in univariable and multivariable linear K+ estimators. Agreement between actual and estimated K+ was confirmed, with averaged errors over patients and time points being 0.000 ± 0.875 mM and 0.046 ± 0.690 mM for stage-specific and patient-specific multivariable K+ estimators, respectively.

CONCLUSION

ECG descriptors of T wave nonlinear dynamics and morphological variability allow noninvasive monitoring of K+ in ESRD patients.

SIGNIFICANCE

ECG markers have the potential to be used for hypo- and hyperkalemia screening in ESRD patients.

Artificial intelligence and cardiology: Current status and perspective: Artificial Intelligence and Cardiology

The development of artificial intelligence (AI) began in the mid-20th century but has been rapidly accelerating in the past decade. Reflecting the development of digital health over the past few years, this trend is also seen in medicine. The field of cardiovascular medicine uses a wide variety and a large amount of biosignals, so there are many situations where AI can contribute. The development of AI is in progress for all aspects of the healthcare system, including the prevention, screening, and treatment of diseases and the prediction of the prognosis. AI is expected to be used to provide specialist-level medical care, even in a situation where medical resources are scarce. However, like other medical devices, the concept and mechanism of AI must be fully understood when used; otherwise, it may be used inappropriately, resulting in detriment to the patient. Therefore, it is important to understand what we need to know as a cardiologist handling AI. This review introduces the basics and principles of AI, then shows how far the current development of AI has come, and finally gives a brief introduction of how to start the AI development for those who want to develop their own AI.

Predictive performance and impact of algorithms in remote monitoring of chronic conditions: A systematic review and meta-analysis

BACKGROUND

The use of telehealth interventions, such as the remote monitoring of patient clinical data (e.g. blood pressure, blood glucose, heart rate, medication use), has been proposed as a strategy to better manage chronic conditions and to reduce the impact on patients and healthcare systems. The use of algorithms for data acquisition, analysis, transmission, communication and visualisation are now common in remote patient monitoring. However, their use and impact on chronic disease management has not been systematically investigated.

OBJECTIVES

To investigate the use, impact, and performance of remote monitoring algorithms across various types of chronic conditions.

METHODS

A literature search of MEDLINE complete, CINHAL complete, and EMBASE was performed using search terms relating to the concepts of remote monitoring, chronic conditions, and data processing algorithms. Comparable outcomes from studies describing the impact on process measures and clinical and patient-reported outcomes were pooled for a summary effect and meta-analyses. A comparison of studies reporting the predictive performance of algorithms was also conducted using the Youden Index.

RESULTS

A total of 89 articles were included in the review. There was no evidence of a positive impact on healthcare utilisation [OR 1.09 (0.90 to 1.31); P = .35] and mortality [OR 0.83 (0.63 to 1.10); P = .208], but there was a positive effect on generic health status [SDM 0.2912 (0.06 to 0.51); P = .010] and diabetes control [SDM -0.53 (-0.74 to -0.33); P < .001; I² = 15.71] (with two of the three diabetes studies being identified as having a high risk of bias). While the majority of impact studies made use of heuristic threshold-based algorithms (n = 27,87%), most performance studies (n = 36, 62%) analysed non-sequential machine learning methods. There was considerable variance in the quality, sample size and performance amongst these studies. Overall, algorithms involved in diagnosis (n = 22, 47%) had superior performance to those involved in predicting a future event (n = 25, 53%). Detection of arrythmia and ischaemia utilising ECG data showed particularly promising results.

CONCLUSION

The performance of data processing algorithms for the diagnosis of a current condition, particularly those related to the detection of arrythmia and ischaemia, is promising. However, there appears to exist minimal testing in experimental studies, with only two included impact studies citing a performance study as support for the intervention algorithm used. Because of the disconnect between performance and impact studies, there is currently limited evidence of the effect of integrating advanced inference algorithms in remote monitoring interventions. If the field of remote patient monitoring is to progress, future impact studies should address this disconnect by evaluating high performance validated algorithms in robust clinical trials.

Development and validation of a deep learning model to screen hypokalemia from electrocardiogram in emergency patients

Background:

A deep learning model (DLM) that enables non-invasive hypokalemia screening from an electrocardiogram (ECG) may improve the detection of this life-threatening condition. This study aimed to develop and evaluate the performance of a DLM for the detection of hypokalemia from the ECGs of emergency patients.

Methods:

We used a total of 9908 ECG data from emergency patients who were admitted at the Second Affiliated Hospital of Nanchang University, Jiangxi, China, from September 2017 to October 2020. The DLM was trained using 12 ECG leads (lead I, II, III, aVR, aVL, aVF, and V₁–₆) to detect patients with serum potassium concentrations <3.5 mmol/L and was validated using retrospective data from the Jiangling branch of the Second Affiliated Hospital of Nanchang University. The blood draw was completed within 10 min before and after the ECG examination, and there was no new or ongoing infusion during this period.

Results:

We used 6904 ECGs and 1726 ECGs as development and internal validation data sets, respectively. In addition, 1278 ECGs from the Jiangling branch of the Second Affiliated Hospital of Nanchang University were used as external validation data sets. Using 12 ECG leads (leads I, II, III, aVR, aVL, aVF, and V₁–₆), the area under the receiver operating characteristic curve (AUC) of the DLM was 0.80 (95% confidence interval [CI]: 0.77–0.82) for the internal validation data set. Using an optimal operating point yielded a sensitivity of 71.4% and a specificity of 77.1%. Using the same 12 ECG leads, the external validation data set resulted in an AUC for the DLM of 0.77 (95% CI: 0.75–0.79). Using an optimal operating point yielded a sensitivity of 70.0% and a specificity of 69.1%.

Conclusions:

In this study, using 12 ECG leads, a DLM detected hypokalemia in emergency patients with an AUC of 0.77 to 0.80. Artificial intelligence could be used to analyze an ECG to quickly screen for hypokalemia.

Deep neural networks learn by using human-selected electrocardiogram features and novel features

AIMS

We sought to investigate whether artificial intelligence (AI) and specifically deep neural networks (NNs) for electrocardiogram (ECG) signal analysis can be explained using human-selected features. We also sought to quantify such explainability and test if the AI model learns features that are similar to a human expert.

METHODS AND RESULTS

We used a set of 100 000 ECGs that were annotated by human explainable features. We applied both linear and non-linear models to predict published ECG AI models output for the detection of patients' age and sex. We further used canonical correlation analysis to quantify the amount of shared information between the NN features and human-selected features. We reconstructed single human-selected ECG features from the unexplained NN features using a simple linear model. We noticed a strong correlation between the simple models and the AI output ( R 2 of 0.49-0.57 for the linear models and R 2 of 0.69-0.70 for the non-linear models). We found that the correlation of the human explainable features with either 13 of the strongest age AI features or 15 of the strongest sex AI features was above 0.85 (for comparison, the first 14 principal components explain 90% of the human feature variance). We linearly reconstructed single human-selected ECG features from the AI features with R 2 up to 0.86.

CONCLUSION

This work shows that NNs for ECG signals extract features in a similar manner to human experts and that they also generate additional novel features that help achieve superior performance.

Deep Learning for ECG Analysis: Benchmarks and Insights from PTB-XL

Electrocardiography (ECG) is a very common, non-invasive diagnostic procedure and its interpretation is increasingly supported by algorithms. The progress in the field of automatic ECG analysis has up to now been hampered by a lack of appropriate datasets for training as well as a lack of well-defined evaluation procedures to ensure comparability of different algorithms. To alleviate these issues, we put forward first benchmarking results for the recently published, freely accessible clinical 12-lead ECG dataset PTB-XL, covering a variety of tasks from different ECG statement prediction tasks to age and sex prediction. Among the investigated deep-learning-based timeseries classification algorithms, we find that convolutional neural networks, in particular resnet- and inception-based architectures, show the strongest performance across all tasks. We find consistent results on the ICBEB2018 challenge ECG dataset and discuss prospects of transfer learning using classifiers pretrained on PTB-XL. These benchmarking results are complemented by deeper insights into the classification algorithm in terms of hidden stratification, model uncertainty and an exploratory interpretability analysis, which provide connecting points for future research on the dataset. Our results emphasize the prospects of deep-learning-based algorithms in the field of ECG analysis, not only in terms of quantitative accuracy but also in terms of clinically equally important further quality metrics such as uncertainty quantification and interpretability. With this resource, we aim to establish the PTB-XL dataset as a resource for structured benchmarking of ECG analysis algorithms and encourage other researchers in the field to join these efforts.

Monitoring blood potassium concentration in hemodialysis patients by quantifying T-wave morphology dynamics

We investigated the ability of time-warping-based ECG-derived markers of T-wave morphology changes in time (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w}$$\end{document}dw) and amplitude (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_a$$\end{document}da), as well as their non-linear components (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${d_w^{{\mathrm{NL}}}}$$\end{document}dwNL and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${d_a^{\mathrm{NL}}}$$\end{document}daNL), and the heart rate corrected counterpart (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w,c}$$\end{document}dw,c), to monitor potassium concentration (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[K^{+}]$$\end{document}[K+]) changes (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta [K^+]$$\end{document}Δ[K+]) in end-stage renal disease (ESRD) patients undergoing hemodialysis (HD). We compared the performance of the proposed time-warping markers, together with other previously proposed \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[K^{+}]$$\end{document}[K+] markers, such as T-wave width (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_w$$\end{document}Tw) and T-wave slope-to-amplitude ratio (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{S/A}$$\end{document}TS/A), when computed from standard ECG leads as well as from principal component analysis (PCA)-based leads. 48-hour ECG recordings and a set of hourly-collected blood samples from 29 ESRD-HD patients were acquired. Values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_w$$\end{document}dw, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_a$$\end{document}da, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${d_w^{\mathrm{NL}}}$$\end{document}dwNL, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${d_a^{\mathrm{NL}}}$$\end{document}daNL and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w,c}$$\end{document}dw,c were calculated by comparing the morphology of the mean warped T-waves (MWTWs) derived at each hour along the HD with that from a reference MWTW, measured at the end of the HD. From the same MWTWs \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_w$$\end{document}Tw and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{S/A}$$\end{document}TS/A were also extracted. Similarly, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta [K^+]$$\end{document}Δ[K+] was calculated as the difference between the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[K^{+}]$$\end{document}[K+] values at each hour and the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[K^{+}]$$\end{document}[K+] reference level at the end of the HD session. We found that \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w}$$\end{document}dw and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w,c}$$\end{document}dw,c showed higher correlation coefficients with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta [K^+]$$\end{document}Δ[K+] than \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_{S/A}$$\end{document}TS/A—Spearman’s (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho$$\end{document}ρ) and Pearson’s (r)—and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_w$$\end{document}Tw—Spearman’s (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho$$\end{document}ρ)—in both SL and PCA approaches being the intra-patient median \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho \ge 0.82$$\end{document}ρ≥0.82 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r \ge 0.87$$\end{document}r≥0.87 in SL and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho \ge 0.82$$\end{document}ρ≥0.82 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r \ge 0.89$$\end{document}r≥0.89 in PCA respectively. Our findings would point at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w}$$\end{document}dw and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{w,c}$$\end{document}dw,c as the most suitable surrogate of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta [K^+]$$\end{document}Δ[K+], suggesting that they could be potentially useful for non-invasive monitoring of ESRD-HD patients in hospital, as well as in ambulatory settings. Therefore, the tracking of T-wave morphology variations by means of time-warping analysis could improve continuous and remote \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[K^{+}]$$\end{document}[K+] monitoring of ESRD-HD patients and flagging risk of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[K^{+}]$$\end{document}[K+]-related cardiovascular events.

Artificial intelligence in cardiovascular medicine

PURPOSE OF REVIEW

Artificial intelligence is a broad set of sophisticated computer-based statistical tools that have become widely available. Cardiovascular medicine with its large data repositories, need for operational efficiency and growing focus on precision care is set to be transformed by artificial intelligence. Applications range from new pathophysiologic discoveries to decision support for individual patient care to optimization of system-wide logistical processes.

RECENT FINDINGS

Machine learning is the dominant form of artificial intelligence wherein complex statistical algorithms 'learn' by deducing patterns in datasets. Supervised machine learning uses classified large data to train an algorithm to accurately predict the outcome, whereas in unsupervised machine learning, the algorithm uncovers mathematical relationships within unclassified data. Artificial multilayered neural networks or deep learning is one of the most successful tools. Artificial intelligence has demonstrated superior efficacy in disease phenomapping, early warning systems, risk prediction, automated processing and interpretation of imaging, and increasing operational efficiency.

SUMMARY

Artificial intelligence demonstrates the ability to learn through assimilation of large datasets to unravel complex relationships, discover prior unfound pathophysiological states and develop predictive models. Artificial intelligence needs widespread exploration and adoption for large-scale implementation in cardiovascular practice.

Artificial intelligence-enhanced electrocardiography in cardiovascular disease management

The application of artificial intelligence (AI) to the electrocardiogram (ECG), a ubiquitous and standardized test, is an example of the ongoing transformative effect of AI on cardiovascular medicine. Although the ECG has long offered valuable insights into cardiac and non-cardiac health and disease, its interpretation requires considerable human expertise. Advanced AI methods, such as deep-learning convolutional neural networks, have enabled rapid, human-like interpretation of the ECG, while signals and patterns largely unrecognizable to human interpreters can be detected by multilayer AI networks with precision, making the ECG a powerful, non-invasive biomarker. Large sets of digital ECGs linked to rich clinical data have been used to develop AI models for the detection of left ventricular dysfunction, silent (previously undocumented and asymptomatic) atrial fibrillation and hypertrophic cardiomyopathy, as well as the determination of a person's age, sex and race, among other phenotypes. The clinical and population-level implications of AI-based ECG phenotyping continue to emerge, particularly with the rapid rise in the availability of mobile and wearable ECG technologies. In this Review, we summarize the current and future state of the AI-enhanced ECG in the detection of cardiovascular disease in at-risk populations, discuss its implications for clinical decision-making in patients with cardiovascular disease and critically appraise potential limitations and unknowns.

Characterization of T Wave Amplitude, Duration and Morphology Changes During Hemodialysis: Relationship With Serum Electrolyte Levels and Heart Rate

OBJECTIVE

Chronic kidney disease affects more than 10% of the world population. Changes in serum ion concentrations increase the risk for ventricular arrhythmias and sudden cardiac death, particularly in end-stage renal disease (ESRD) patients. We characterized how T wave amplitude, duration and morphology descriptors change with variations in serum levels of potassium and calcium and in heart rate, both in ESRD patients and in simulated ventricular fibers.

METHODS

Electrocardiogram (ECG) recordings from twenty ESRD patients undergoing hemodialysis (HD) and pseudo-ECGs (pECGs) calculated from twenty-two simulated ventricular fibers at varying transmural heterogeneity levels were processed to quantify T wave width ( T_w), T wave slope-to-amplitude ratio ([Formula: see text]) and four indices of T wave morphological variability based on time warping ( d_w, [Formula: see text], d_a and [Formula: see text]). Serum potassium and calcium levels and heart rate were measured along HD.

RESULTS

[Formula: see text] was the marker most strongly correlated with serum potassium, dw with calcium and da with heart rate, after correction for covariates. Median values of partial correlation coefficients were 0.75, -0.74 and -0.90, respectively. For all analyzed T wave descriptors, high inter-patient variability was observed in the pattern of such relationships. This variability, accentuated during the first HD time points, was reproduced in the simulations and shown to be influenced by differences in transmural heterogeneity.

CONCLUSION

Changes in serum potassium and calcium levels and in heart rate strongly affect T wave descriptors, particularly those quantifying morphological variability.

SIGNIFICANCE

ECG markers have the potential to be used for monitoring serum ion concentrations in ESRD patients.

Quantification and classification of potassium and calcium disorders with the electrocardiogram: What do clinical studies, modeling, and reconstruction tell us?

Diseases caused by alterations of ionic concentrations are frequently observed challenges and play an important role in clinical practice. The clinically established method for the diagnosis of electrolyte concentration imbalance is blood tests. A rapid and non-invasive point-of-care method is yet needed. The electrocardiogram (ECG) could meet this need and becomes an established diagnostic tool allowing home monitoring of the electrolyte concentration also by wearable devices. In this review, we present the current state of potassium and calcium concentration monitoring using the ECG and summarize results from previous work. Selected clinical studies are presented, supporting or questioning the use of the ECG for the monitoring of electrolyte concentration imbalances. Differences in the findings from automatic monitoring studies are discussed, and current studies utilizing machine learning are presented demonstrating the potential of the deep learning approach. Furthermore, we demonstrate the potential of computational modeling approaches to gain insight into the mechanisms of relevant clinical findings and as a tool to obtain synthetic data for methodical improvements in monitoring approaches.

Artificial Intelligence-Enabled ECG: a Modern Lens on an Old Technology

PURPOSE OF REVIEW

To (i) review the concept of artificial intelligence (AI); (ii) summarize recent developments in artificial intelligence-enabled electrocardiogram (AI-ECG); (iii) address notable inherent limitations and challenges of AI-ECG; and (iv) discuss the future direction of the field.

RECENT FINDINGS

Advancements in machine learning and computing methods have led to application of AI-ECG and potential new applications to patient care. Further study is needed to verify previous findings in diverse populations as well as begin to confront the limitations needed for clinical implementation. Nearly one century after the Nobel Prize was awarded to Willem Einthoven for demonstrating that an electrocardiogram (ECG) could record the electrical signature of the heart, the ECG remains one of the most important diagnostic tests in modern medicine. We now stand at the edge of true ECG innovation. Simultaneous advancements in computing power, wireless technology, digitized data availability, and machine learning have led to the birth of AI-ECG algorithms with novel capabilities and real potential for clinical application. AI has the potential to improve diagnostic accuracy and efficiency by providing fully automated, unbiased, and unambiguous ECG analysis along with promising new findings that may unlock new value in the ECG. These breakthroughs may cause a paradigm shift in clinical workflow as well as patient monitoring and management.

Development and Validation of a Deep-Learning Model to Screen for Hyperkalemia From the Electrocardiogram

Importance

For patients with chronic kidney disease (CKD), hyperkalemia is common, associated with fatal arrhythmias, and often asymptomatic, while guideline-directed monitoring of serum potassium is underused. A deep-learning model that enables noninvasive hyperkalemia screening from the electrocardiogram (ECG) may improve detection of this life-threatening condition.

Objective

To evaluate the performance of a deep-learning model in detection of hyperkalemia from the ECG in patients with CKD.

Design, Setting, and Participants

A deep convolutional neural network (DNN) was trained using 1 576 581 ECGs from 449 380 patients seen at Mayo Clinic, Rochester, Minnesota, from 1994 to 2017. The DNN was trained using 2 (leads I and II) or 4 (leads I, II, V3, and V5) ECG leads to detect serum potassium levels of 5.5 mEq/L or less (to convert to millimoles per liter, multiply by 1) and was validated using retrospective data from the Mayo Clinic in Minnesota, Florida, and Arizona. The validation included 61 965 patients with stage 3 or greater CKD. Each patient had a serum potassium count drawn within 4 hours after their ECG was recorded. Data were analyzed between April 12, 2018, and June 25, 2018.

Exposures

Use of a deep-learning model.

Main Outcomes and Measures

Area under the receiver operating characteristic curve (AUC) and sensitivity and specificity, with serum potassium level as the reference standard. The model was evaluated at 2 operating points, 1 for equal specificity and sensitivity and another for high (90%) sensitivity.

Results

Of the total 1 638 546 ECGs, 908 000 (55%) were from men. The prevalence of hyperkalemia in the 3 validation data sets ranged from 2.6% (n = 1282 of 50 099; Minnesota) to 4.8% (n = 287 of 6011; Florida). Using ECG leads I and II, the AUC of the deep-learning model was 0.883 (95% CI, 0.873-0.893) for Minnesota, 0.860 (95% CI, 0.837-0.883) for Florida, and 0.853 (95% CI, 0.830-0.877) for Arizona. Using a 90% sensitivity operating point, the sensitivity was 90.2% (95% CI, 88.4%-91.7%) and specificity was 63.2% (95% CI, 62.7%-63.6%) for Minnesota; the sensitivity was 91.3% (95% CI, 87.4%-94.3%) and specificity was 54.7% (95% CI, 53.4%-56.0%) for Florida; and the sensitivity was 88.9% (95% CI, 84.5%-92.4%) and specificity was 55.0% (95% CI, 53.7%-56.3%) for Arizona.

Conclusions and Relevance

In this study, using only 2 ECG leads, a deep-learning model detected hyperkalemia in patients with renal disease with an AUC of 0.853 to 0.883. The application of artificial intelligence to the ECG may enable screening for hyperkalemia. Prospective studies are warranted.

Artificial Intelligence in Cardiology: Present and Future

Artificial intelligence (AI) is a nontechnical, popular term that refers to machine learning of various types but most often to deep neural networks. Cardiology is at the forefront of AI in medicine. For this review, we searched PubMed and MEDLINE databases with no date restriction using search terms related to AI and cardiology. Articles were selected for inclusion on the basis of relevance. We highlight the major achievements in recent years in nearly all areas of cardiology and underscore the mounting evidence suggesting how AI will take center stage in the field. Artificial intelligence requires a close collaboration among computer scientists, clinical investigators, clinicians, and other users in order to identify the most relevant problems to be solved. Best practices in the generation and implementation of AI include the selection of ideal data sources, taking into account common challenges during the interpretation, validation, and generalizability of findings, and addressing safety and ethical concerns before final implementation. The future of AI in cardiology and in medicine in general is bright as the collaboration between investigators and clinicians continues to excel.

A Deep-Learning Algorithm (ECG12Net) for Detecting Hypokalemia and Hyperkalemia by Electrocardiography: Algorithm Development

Background

The detection of dyskalemias—hypokalemia and hyperkalemia—currently depends on laboratory tests. Since cardiac tissue is very sensitive to dyskalemia, electrocardiography (ECG) may be able to uncover clinically important dyskalemias before laboratory results.

Objective

Our study aimed to develop a deep-learning model, ECG12Net, to detect dyskalemias based on ECG presentations and to evaluate the logic and performance of this model.

Methods

Spanning from May 2011 to December 2016, 66,321 ECG records with corresponding serum potassium (K⁺) concentrations were obtained from 40,180 patients admitted to the emergency department. ECG12Net is an 82-layer convolutional neural network that estimates serum K⁺ concentration. Six clinicians—three emergency physicians and three cardiologists—participated in human-machine competition. Sensitivity, specificity, and balance accuracy were used to evaluate the performance of ECG12Net with that of these physicians.

Results

In a human-machine competition including 300 ECGs of different serum K+ concentrations, the area under the curve for detecting hypokalemia and hyperkalemia with ECG12Net was 0.926 and 0.958, respectively, which was significantly better than that of our best clinicians. Moreover, in detecting hypokalemia and hyperkalemia, the sensitivities were 96.7% and 83.3%, respectively, and the specificities were 93.3% and 97.8%, respectively. In a test set including 13,222 ECGs, ECG12Net had a similar performance in terms of sensitivity for severe hypokalemia (95.6%) and severe hyperkalemia (84.5%), with a mean absolute error of 0.531. The specificities for detecting hypokalemia and hyperkalemia were 81.6% and 96.0%, respectively.

Conclusions

A deep-learning model based on a 12-lead ECG may help physicians promptly recognize severe dyskalemias and thereby potentially reduce cardiac events.

Detection and Classification of Cardiac Arrhythmias by a Challenge-Best Deep Learning Neural Network Model

Electrocardiograms (ECGs) are widely used to clinically detect cardiac arrhythmias (CAs). They are also being used to develop computer-assisted methods for heart disease diagnosis. We have developed a convolution neural network model to detect and classify CAs, using a large 12-lead ECG dataset (6,877 recordings) provided by the China Physiological Signal Challenge (CPSC) 2018. Our model, which was ranked first in the challenge competition, achieved a median overall F1-score of 0.84 for the nine-type CA classification of CPSC2018's hidden test set of 2,954 ECG recordings. Further analysis showed that concurrent CAs were adequately predictive for 476 patients with multiple types of CA diagnoses in the dataset. Using only single-lead data yielded a performance that was only slightly worse than using the full 12-lead data, with leads aVR and V1 being the most prominent. We extensively consider these results in the context of their agreement with and relevance to clinical observations.

•

Accurate AI diagnosis of cardiac arrhythmia on ECG data from 11 hospitals

•

Capable of diagnosing concurrent cardiac arrhythmias

•

An ensemble model combining 12- and 1-lead models ranked first in CPSC2018

•

aVR and V1 found to be the best-performing single leads

Age and Sex Estimation Using Artificial Intelligence From Standard 12-Lead ECGs

Supplemental Digital Content is available in the text.

Sex and age have long been known to affect the ECG. Several biologic variables and anatomic factors may contribute to sex and age-related differences on the ECG. We hypothesized that a convolutional neural network (CNN) could be trained through a process called deep learning to predict a person’s age and self-reported sex using only 12-lead ECG signals. We further hypothesized that discrepancies between CNN-predicted age and chronological age may serve as a physiological measure of health.

Can physicians detect hyperkalemia based on the electrocardiogram?

OBJECTIVE

Although there is no consensus on how to use an electrocardiogram (ECG) in patients with hyperkalemia, physicians often obtain it in the acute setting when diagnosing and treating hyperkalemia. The objective of this study is to evaluate if physicians are able to detect hyperkalemia based on the ECG.

METHODS

The study was conducted at a large county hospital with a population of end stage renal disease (ESRD) patients who received hemodialysis (HD) solely on an emergent basis. Five hundred twenty eight ECGs from ESRD patients were evaluated. The prevalence of hyperkalemia was approximately 60% in this cohort, with at least half of them in the severe hyperkalemia range (K ≥ 6.5 mEq/L).

RESULTS

The mean sensitivity and specificity of the emergency physicians detecting hyperkalemia were 0.19 (± 0.16) and 0.97(± 0.04) respectively. The mean positive predictive value of evaluators for detecting hyperkalemia was 0.92 (±0.13) and the mean negative predictive value was 0.46 (± 0.05). In severe hyperkalemia (K ≥ 6.5 mEq/L), the mean sensitivity improved to 0.29 (± 0.20), while specificity decreased to 0.95 (±0.07).

CONCLUSION

An ECG is not a sensitive method of detecting hyperkalemia and should not be relied upon to rule it out. However, the ECG has a high specificity for detecting hyperkalemia and could be used as a rule in test.

Quantitative Evaluation of the Relationship between T-Wave-Based Features and Serum Potassium Level in Real-World Clinical Practice

BACKGROUND

Proper management of hyperkalemia that leads to fatal cardiac arrhythmia has become more important because of the increased prevalence of hyperkalemia-prone diseases. Although T-wave changes in hyperkalemia are well known, their usefulness is debatable. We evaluated how well T-wave-based features of electrocardiograms (ECGs) are correlated with estimated serum potassium levels using ECG data from real-world clinical practice.

METHODS

We collected ECGs from a local ECG repository (MUSE™) from 1994 to 2017 and extracted the ECG waveforms. Of about 1 million reports, 124,238 were conducted within 5 minutes before or after blood collection for serum potassium estimation. We randomly selected 500 ECGs and two evaluators measured the amplitude (T-amp) and right slope of the T-wave (T-right slope) on five lead waveforms (V3, V4, V5, V6, and II). Linear correlations of T-amp, T-right slope, and their normalized feature (T-norm) with serum potassium levels were evaluated using Pearson correlation coefficient analysis.

RESULTS

Pearson correlation coefficients for T-wave-based features with serum potassium between the two evaluators were 0.99 for T-amp and 0.97 for T-right slope. The coefficient for the association between T-amp, T-right slope, and T-norm, and serum potassium ranged from -0.22 to 0.02. In the normal ECG subgroup (normal ECG or otherwise normal ECG), there was no correlation between T-wave-based features and serum potassium level.

CONCLUSIONS

T-wave-based features were not correlated with serum potassium level, and their use in real clinical practice is currently limited.

Errors of Classification With Potassium Blood Testing: The Variability and Repeatability of Critical Clinical Tests

OBJECTIVE

To understand the performance of a currently used clinical blood test with regard to the frequency and size of variation of the results.

PATIENTS AND METHODS

From November 29, 2012, through November 29, 2013, patients were recruited at 65 sites as part of a previously reported clinical trial (ClinicalTrials.gov Identifier: NCT01737697). Eligible outpatients who had been fasting for at least 8 hours underwent venous phlebotomy at baseline, 30 minutes, and 60 minutes to measure plasma potassium levels in whole blood using a point-of-care device (i-STAT, Abbott Laboratories). We analyzed the results to assess their variability and frequency of pseudohyperkalemia and pseudonormokalemia.

RESULTS

A total of 1170 patients were included in this study. Absolute differences between pairs of measurements from different time points ranged from 0 to 2.5 mmol/L, with a mean difference of 0.26 mmol/L. The mean percentage differences were approximately 5% with an SD of 5%. Approximately 12% of differences between repeated fasting potassium blood test results were above 0.5 mmol/L (33% of the normal range), and 20% of patients (234) had at least one difference greater than 0.5 mmol/L. In 44.0% of the patients with a hyperkalemic average value (true hyperkalemia) (302 of 686), at least one blood test result was in the normal range (pseudonormokalemia), and in 30.2% of the patients with a normal average value (146 of 484), at least one blood test result was elevated (pseudohyperkalemia).

CONCLUSION

Expected variability and errors exist with potassium blood tests, even when conditions are optimized. Pseudohyperkalemia and pseudonormokalemia are common, indicating a need for thoughtful clinical interpretation of unexpected test results.

System for Collecting Biosignal Data from Multiple Patient Monitoring Systems

Objectives

Biosignal data include important physiological information. For that reason, many devices and systems have been developed, but there has not been enough consideration of how to collect and integrate raw data from multiple systems. To overcome this limitation, we have developed a system for collecting and integrating biosignal data from two patient monitoring systems.

Methods

We developed an interface to extract biosignal data from Nihon Kohden and Philips monitoring systems. The Nihon Kohden system has a central server for the temporary storage of raw waveform data, which can be requested using the HL7 protocol. However, the Philips system used in our hospital cannot save raw waveform data. Therefore, our system was connected to monitoring devices using the RS232 protocol. After collection, the data were transformed and stored in a unified format.

Results

From September 2016 to August 2017, we collected approximately 117 patient-years of waveform data from 1,268 patients in 79 beds of five intensive care units. Because the two systems use the same data storage format, the application software could be run without compatibility issues.

Conclusions

Our system collects biosignal data from different systems in a unified format. The data collected by the system can be used to develop algorithms or applications without the need to consider the source of the data.

Noninvasive blood potassium measurement using signal-processed, single-lead ecg acquired from a handheld smartphone

OBJECTIVE

We have previously used a 12-lead, signal-processed ECG to calculate blood potassium levels. We now assess the feasibility of doing so with a smartphone-enabled single lead, to permit remote monitoring.

PATIENTS AND METHODS

Twenty-one hemodialysis patients held a smartphone equipped with inexpensive FDA-approved electrodes for three 2min intervals during hemodialysis. Individualized potassium estimation models were generated for each patient. ECG-calculated potassium values were compared to blood potassium results at subsequent visits to evaluate the accuracy of the potassium estimation models.

RESULTS

The mean absolute error between the estimated potassium and blood potassium 0.38±0.32 mEq/L (9% of average potassium level) decreasing to 0.6 mEq/L using predictors of poor signal.

CONCLUSIONS

A single-lead ECG acquired using electrodes attached to a smartphone device can be processed to calculate the serum potassium with an error of 9% in patients undergoing hemodialysis.

SUMMARY

A single-lead ECG acquired using electrodes attached to a smartphone can be processed to calculate the serum potassium in patients undergoing hemodialysis remotely.

Noninvasive quantification of blood potassium concentration from ECG in hemodialysis patients

Blood potassium concentration ([K⁺]) influences the electrocardiogram (ECG), particularly T-wave morphology. We developed a new method to quantify [K⁺] from T-wave analysis and tested its clinical applicability on data from dialysis patients, in whom [K⁺] varies significantly during the therapy. To elucidate the mechanism linking [K⁺] and T-wave, we also analysed data from long QT syndrome type 2 (LQT2) patients, testing the hypothesis that our method would have underestimated [K⁺] in these patients. Moreover, a computational model was used to explore the physiological processes underlying our estimator at the cellular level. We analysed 12-lead ECGs from 45 haemodialysis and 12 LQT2 patients. T-wave amplitude and downslope were calculated from the first two eigenleads. The T-wave slope-to-amplitude ratio (T_S/A) was used as starting point for an ECG-based [K⁺] estimate (K_ECG). Leave-one-out cross-validation was performed. Agreement between K_ECG and reference [K⁺] from blood samples was promising (error: −0.09 ± 0.59 mM, absolute error: 0.46 ± 0.39 mM). The analysis on LQT2 patients, also supported by the outcome of computational analysis, reinforces our interpretation that, at the cellular level, delayed-rectifier potassium current is a main contributor of K_ECG correlation to blood [K⁺]. Following a comprehensive validation, this method could be effectively applied to monitor patients at risk for hyper/hypokalemia.