ECG for laboratory values:A comprehensive physiological and technical analysis

Walentin Palanchuk

cs.LGOct 20, 2026

ECG for laboratory values:A comprehensive physiological and technical analysis

Authors: Walentin Palanchuk

Organizations: Carl von Ossietzky Universität Oldenburg Fach-Bachelor Informatik

Abstract

Laboratory measurements are essential for clinical decision making but typically require invasive blood sampling. Recent research suggests that ECG recordings may contain information beyond cardiac rhythm and provide information related to laboratory values. This thesis investigates NT-proBNP prediction from ECG recordings and patient metadata using a multimodal deep learning framework, while considering the physiological plausibility of ECG-based prediction. It also compares a 1-lead configuration using lead II with the corresponding 12-lead configuration. Additional experiments on glucose, hemoglobin, and potassium are presented in the appendix. Regression, binary classification, and three-class classification experiments were conducted using MIMIC-IV, with the three-class task based on clinically established thresholds. The 12-lead configuration consistently achieved better performance. Regression reached a mean absolute error of 534.98 pg/mL for 1-lead and 514.27 pg/mL for 12-lead. Binary classification achieved AUROCs of 0.849 and 0.876, respectively, while three-class classification achieved macro-averaged AUROCs of 0.738 and 0.752. The results indicate that precise NT-proBNP regression remains challenging, whereas broader concentration ranges can be classified more effectively. The findings are consistent with the physiological hypothesis that ECG characteristics and NT-proBNP concentrations reflect shared cardiovascular processes. ECG-based NT-proBNP prediction may therefore provide complementary information, particularly for classifying broader concentration ranges. However, laboratory measurements remain necessary, and validation on independent datasets is required.

Figures & tables

Task	Thresholds [pg/mL]	Leads	Loss
Regression	–	1-lead	L1
Regression	–	12-lead	L1
Binary classification	300	1-lead	Weighted focal CE
Binary classification	300	12-lead	Weighted focal CE
Three-class classification	400, 2000	1-lead	Weighted focal CE
Three-class classification	400, 2000	12-lead	Weighted focal CE

Table 1: Overview of the primary NT-proBNP experiments.

Figure 1: Schematic overview of the temporal matching of ECG and laboratory values.

Feature	Source	Processing
Age	MIMIC-IV	Calculated at ECG time
Sex	MIMIC-IV	Binary encoding
RR interval	MIMIC-IV-ECG	Machine measurement
P wave onset	MIMIC-IV-ECG	Machine measurement
P wave end	MIMIC-IV-ECG	Machine measurement
QRS onset	MIMIC-IV-ECG	Machine measurement

Table 2: Tabular input features

Figure 2: Overview of the pipeline for the laboratory value prediction.

Parameter	Value
Epochs	20
Batch size	32
Optimizer	AdamW
Learning rate	0.001
Weight decay	0.001
Precision	32-bit

Table 3: Training configuration for the primary NT-proBNP experiments.

Task	ECG recordings	ECG segments
Regression	910	3640
Classification	1435	5740

Table 4: Test split size for the primary NT-proBNP experiments.

Task	Class	ECG recordings	ECG segments
Binary	$<300$ pg/mL	307	1228
	$\geq 300$ pg/mL	1128	4512
Three-class	$<400$ pg/mL	381	1524
	$400$ – $2000$ pg/mL	410	1640
	$>2000$ pg/mL	644	2576

Table 5: Test split class distribution for the NT-proBNP classification experiments.

Feature	Regression n (%)	Classification n (%)
BMI	229 (25.2%)	418 (29.1%)
Height	367 (40.3%)	653 (45.5%)
Weight	187 (20.5%)	325 (22.6%)
Temperature	628 (69.0%)	1057 (73.7%)
Heart rate	470 (51.6%)	776 (54.1%)
Respiratory rate	472 (51.9%)	781 (54.4%)

Table 6: Missing metadata counts in the test splits for the primary NT-proBNP experiments.

ECG configuration	MAE [pg/mL]	95% CI [pg/mL]
1-lead	534.98	501.16–571.62
12-lead	514.27	482.83–546.46

Table 7: Mean absolute error with 95% confidence intervals for NT-proBNP prediction.

Panel of Figure 3 — Figure 3: Bland–Altman analysis of predicted and reference NT-proBNP concentrations for the 1-lead and 12-lead configurations. The solid line represents the mean difference and the dashed lines indicate the 95% limits of agreement.

Panel of Figure 4 — Figure 4: Passing–Bablok regression of predicted and reference NT-proBNP concentrations for the 1-lead and 12-lead configurations. The solid line represents the Passing–Bablok regression fit, the dashed line represents the identity, and the shaded area represents the 95% confidence interval.

ECG configuration	AUROC	95% CI	Accuracy	Sensitivity	F1 score
1-lead	0.849	0.839–0.860	0.770	0.784	0.843
12-lead	0.876	0.866–0.886	0.799	0.798	0.862

Table 8: Binary classification performance for NT-proBNP prediction.

Panel of Figure 5 — Figure 5: Confusion matrices for the binary NT-proBNP classification task using the 1-lead and 12-lead ECG configurations.

ECG configuration	AUROC	95% CI	Accuracy	Sensitivity	F1 score
1-lead	0.738	0.729–0.747	0.551	0.527	0.524
12-lead	0.752	0.743–0.761	0.578	0.565	0.539

Table 9: Three-class classification performance for NT-proBNP prediction.

Panel of Figure 6 — Figure 6: Confusion matrices for the three-class NT-proBNP classification task using the 1-lead and 12-lead ECG configurations.

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

Loss configuration	Macro AUROC	95% CI	Accuracy	Sensitivity	F1 score
Cross entropy	0.714	0.705–0.722	0.544	0.413	0.367
Weighted cross entropy	0.726	0.717–0.734	0.471	0.413	0.398
Weighted focal cross entropy	0.709	0.700–0.717	0.478	0.435	0.421

Appendix

Table 10: Predictive results of the NYHA inspired NT-proBNP classification experiments.

Panel of Figure 7 — Figure 7: Confusion matrices for the NYHA inspired NT-proBNP classifications

Target scaling	MAE [pg/mL]	95% CI [pg/mL]
MinMax	2390.46	2314.93–2460.34
Standard	2317.63	2242.62–2396.17

Appendix

Table 11: Predictive results of the additional NT-proBNP regression configurations.

Panel of Figure 8 — Figure 8: Bland–Altman analysis of predicted and reference NT-proBNP concentrations for the MinMax and standard scaling configurations. The solid line represents the mean difference and the dashed lines indicate the 95% limits of agreement.

Panel of Figure 9 — Figure 9: Passing–Bablok regression analysis of predicted and reference NT-proBNP concentrations for the MinMax and standard scaling configurations. The solid line represents the Passing–Bablok regression fit, the dashed line represents the identity, and the shaded area represents the 95% confidence interval.

ECG configuration	MAE [g/dL]	95% CI [g/dL]
1-lead	1.6008	1.5585–1.6441
12-lead	1.5991	1.5562–1.6413

Appendix

Table 12: Mean absolute error with 95% confidence intervals for hemoglobin prediction.

Panel of Figure 10 — Figure 10: Bland–Altman analysis of predicted and reference hemoglobin concentrations for the 1-lead and 12-lead ECG configurations. The solid line represents the mean difference and the dashed lines indicate the 95% limits of agreement.

Panel of Figure 11 — Figure 11: Passing–Bablok regression analysis of predicted and reference hemoglobin concentrations for the 1-lead and 12-lead ECG configurations. The solid line represents the Passing–Bablok regression fit, the dashed line represents the identity, and the shaded area represents the 95% confidence interval.

Loss configuration	MAE [mg/dL]	95% CI [mg/dL]
L1 loss	34.26	33.81–34.69
Huber loss	34.32	33.87–34.81
Gaussian negative log-likelihood	37.75	37.35–38.14
Weighted Clarke error grid loss	48.40	48.04–48.72

Appendix

Table 13: Regression configurations for glucose prediction.

Figure 12: Bland–Altman analysis of predicted and reference glucose concentrations using the

\pm

5 minute temporal window. The solid line represents the mean difference and the dashed lines indicate the 95% limits of agreement.

Figure 13: Passing–Bablok regression analysis for glucose prediction.

ECG configuration	Macro AUROC	95% CI	Accuracy	Sensitivity	F1 score
1-lead	0.632	0.610–0.652	0.554	0.420	0.338

Appendix

Table 14: Three-class classification performance for glucose prediction.

Figure 14: Confusion matrix for the three-class glucose classification experiment.

Figure 15: Bland–Altman analysis of predicted and reference potassium concentrations using the 1-lead ECG configuration. The solid line represents the mean difference and the dashed lines indicate the 95% limits of agreement.

Figure 16: Passing–Bablok regression analysis for potassium prediction.

ECG configuration	Macro AUROC	95% CI	Accuracy	Sensitivity	F1 score
1-lead	0.629	0.617–0.642	0.488	0.444	0.356

Appendix

Table 15: Three-class classification performance for potassium prediction.

Figure 17: Confusion matrix for the three-class potassium classification experiment.

Clinical Prediction

Clinical prediction uses patient records, signals, and images to estimate diagnoses, risks, outcomes, and treatment responses for care.

See all 207 papers

Jul 13, 2026

NeuroECG: ECGFounder-Based Deep ECG Representation for EEG-Free Neurological Prognostication After Cardiac Arrest

Neurological prognostication after cardiac arrest commonly relies on electroencephalography (EEG). However, EEG demands high clinical resources. Bedside electrocardiography (ECG) is standard and low-cost. Yet, its value for predicting neurological outcomes remains underexplored. In this study, we propose NeuroECG, an ECGFounder-based deep representation framework for EEG-free auxiliary prognostication. NeuroECG adapts a pretrained ECG foundation model via task-specific fine-tuning. We implement a gradual unfreezing strategy on single-channel bedside monitoring ECG. Multiple ECG segments per patient are encoded into segment-level deep features. These embeddings are aggregated via quantile pooling (q = 0.24) and compressed using principal component analysis (PCA). Experiments on 412 ECG-available patients from the multicenter I-CARE database show that the adapted ECGFounder backbone achieves the best performance among ECG-only backbone baselines, with a test AUROC of 0.7333. We further combine the learned deep ECG representation with static clinical covariates. The proposed NeuroECG model achieves a test AUROC of 0.8077 and an AUPRC of 0.8970. These results support deep bedside ECG representations as a useful source of auxiliary prognostic information. Their integration with static clinical covariates improves prediction in an EEG-free setting. The source code is available at https://github.com/goddream66/NeuroECG

Jiajun Gao, Yi Zhao, Chenyang Xu +2

School of Economics and Management Xidian University Xi’an, China · School of Mechano-Electronic Engineering Xidian University Xi’an, China · School of Cyber Engineering Xidian University Xi’an, China +1

Representation Learning Healthcare Electrocardiography Clinical Outcome Prediction Electrocardiogram Representation Learning Electrocardiogram Foundation Models+1

Ask PDF

Jul 26, 2026

Blood Pressure Estimation from PPG: A Comparative Study of Direct and ECG-Mediated Deep Learning Pipelines

Continuous cuffless blood pressure (BP) monitoring is essential for connected health systems and wearable devices, enabling early detection, longitudinal tracking, and personalized management of cardiovascular disease. Many prior approaches attempt to estimate BP indirectly by reconstructing electrocardiography (ECG) from photoplethysmography (PPG), assuming ECG provides a stronger physiological link to BP. However, ECG sensing is less accessible in wearable settings and may introduce unnecessary complexity. In this work, we first perform a large-scale physiological correlation analysis on the MIMIC-III waveform database, revealing that PPG exhibits substantially stronger coupling with arterial blood pressure (ABP) (

|r|=0.247

p<0.001

) than ECG does (

r=0.018

p=0.187

), challenging the assumption that ECG provides a superior intermediate representation. Motivated by this insight, we conduct a systematic comparison between direct PPG-to-BP prediction and ECG-mediated pipelines using multiple state-of-the-art deep learning models. Across 1.74M segments from 3,127 patients, direct PPG-to-BP prediction achieves British Hypertension Society Grade A performance (

\mathrm{MAE}_{\mathrm{SBP}} = 4.82 mmHg

\mathrm{MAE}_{\mathrm{DBP}} = 4.31 mmHg

), outperforming all ECG-mediated approaches, which achieve only Grade B accuracy. Our findings suggest that accurate continuous BP monitoring can be achieved directly from wearable PPG signals, enabling simpler, more efficient pipelines for real-world connected health systems.

Bo Wu, Haoling Wang, Zhuodiao Kuang +1

Department of Informatics and Networked Systems University of Pittsburgh Pittsburgh, USA · Department of Biostatistics and Health Data Science University of Pittsburgh Pittsburgh, USA · Language Technologies Institute Carnegie Mellon University Pittsburgh, USA

Photoplethysmography Wearable Health Monitoring Cuffless Blood Pressure Estimation

Ask PDF

Mar 9, 2026

Echo2ECG: Enhancing ECG Representations with Cardiac Morphology from Multi-View Echos

Electrocardiography (ECG) is a low-cost, widely used modality for diagnosing electrical abnormalities like atrial fibrillation by capturing the heart's electrical activity. However, it cannot directly measure cardiac morphological phenotypes, such as left ventricular ejection fraction (LVEF), which typically require echocardiography (Echo). Predicting these phenotypes from ECG would enable early, accessible health screening. Existing self-supervised methods suffer from a representational mismatch by aligning ECGs to single-view Echos, which only capture local, spatially restricted anatomical snapshots. To address this, we propose Echo2ECG, a multimodal self-supervised learning framework that enriches ECG representations with the heart's morphological structure captured in multi-view Echos. We evaluate Echo2ECG as an ECG feature extractor on two clinically relevant tasks that fundamentally require morphological information: (1) classification of structural cardiac phenotypes across three datasets, and (2) retrieval of Echo studies with similar morphological characteristics using ECG queries. Our extracted ECG representations consistently outperform those of state-of-the-art unimodal and multimodal baselines across both tasks, despite being 18x smaller than the largest baseline. These results demonstrate that Echo2ECG is a robust, powerful ECG feature extractor. Our code is accessible at https://github.com/michelleespranita/Echo2ECG.

Michelle Espranita Liman, Özgün Turgut, Alexander Müller +3

Chair for AI in Healthcare and Medicine, Technical University of Munich (TUM) and TUM University Hospital, Germany · Department of Cardiology, TUM University Hospital, Germany · Department of Computing, Imperial College London, UK +1

Electrocardiogram Representation Learning Multimodal Learning Electrocardiogram Classification Self-Supervised Time Series Representation Learning Cross-Modal Retrieval

Ask PDF

Abstract

Clinical Prediction

Explore similar work

NeuroECG: ECGFounder-Based Deep ECG Representation for EEG-Free Neurological Prognostication After Cardiac Arrest

Blood Pressure Estimation from PPG: A Comparative Study of Direct and ECG-Mediated Deep Learning Pipelines

Echo2ECG: Enhancing ECG Representations with Cardiac Morphology from Multi-View Echos