ECG-Based Electrolyte Prediction: Evaluating Regression and Probabilistic Methods

Philipp Von Bachmann,Daniel Gedon,Fredrik K. Gustafsson,Antônio H. Ribeiro,Erik Lampa,Stefan Gustafsson,Johan Sundström,Thomas B. Schön

2022-12-22

Abstract:Objective: Imbalances of the electrolyte concentration levels in the body can lead to catastrophic consequences, but accurate and accessible measurements could improve patient outcomes. While blood tests provide accurate measurements, they are invasive and the laboratory analysis can be slow or inaccessible. In contrast, an electrocardiogram (ECG) is a widely adopted tool which is quick and simple to acquire. However, the problem of estimating continuous electrolyte concentrations directly from ECGs is not well-studied. We therefore investigate if regression methods can be used for accurate ECG-based prediction of electrolyte concentrations. Methods: We explore the use of deep neural networks (DNNs) for this task. We analyze the regression performance across four electrolytes, utilizing a novel dataset containing over 290000 ECGs. For improved understanding, we also study the full spectrum from continuous predictions to binary classification of extreme concentration levels. To enhance clinical usefulness, we finally extend to a probabilistic regression approach and evaluate different uncertainty estimates. Results: We find that the performance varies significantly between different electrolytes, which is clinically justified in the interplay of electrolytes and their manifestation in the ECG. We also compare the regression accuracy with that of traditional machine learning models, demonstrating superior performance of DNNs. Conclusion: Discretization can lead to good classification performance, but does not help solve the original problem of predicting continuous concentration levels. While probabilistic regression demonstrates potential practical usefulness, the uncertainty estimates are not particularly well-calibrated. Significance: Our study is a first step towards accurate and reliable ECG-based prediction of electrolyte concentration levels.

Signal Processing,Computer Vision and Pattern Recognition,Machine Learning

What problem does this paper attempt to address?

The paper primarily focuses on addressing the issue of predicting electrolyte concentrations through electrocardiograms (ECG). Specifically, the objectives of the article include:

1. **Developing Predictive Models**: The researchers aim to use deep neural networks (DNN) to directly predict continuous concentration levels of electrolytes (such as potassium, calcium, sodium, and creatinine) from ECGs.

2. **Evaluating the Effectiveness of Different Methods**: The article compares the performance of DNN with traditional machine learning models (such as gradient boosting and random forest) on the ECG dataset and explores the effectiveness of regression and probabilistic methods.

3. **Utilizing Large Datasets**: The study uses a new large dataset containing over 290,000 ECG records from adult patients in the emergency departments of Swedish hospitals.

4. **Addressing Practical Clinical Needs**: The article discusses practical applications of ECG-based electrolyte concentration prediction, such as quickly identifying causes of arrhythmias in ambulances or providing critical care in areas lacking specialized medical facilities.

5. **Uncertainty Estimation**: To enhance the practical utility of the predictions, the research also explores probabilistic regression methods to provide uncertainty estimates of the predictions, which is crucial for medical applications.

In summary, this paper aims to support clinical decision-making by developing and evaluating ECG-based electrolyte concentration prediction models and exploring how to effectively utilize ECG data for electrolyte monitoring.