Interpretable detection of left ventricular hypertrophy using commercial ECG features and machine learning: a study based on the PTB-XL+ dataset

Q: What is the clinical evidence from this study?

Study design: Observational. Population: Left ventricular hypertrophy (n=11692). Intervention: XGBoost with GE 12SL ECG features vs. Other machine learning models and feature sets (UNIG, ECGDeli). Primary outcome: Area under the receiver-operating-characteristic curve (AUC-ROC) for LVH detection.

Q: What does this research mean for the field?

Machine learning classifiers trained on pre-extracted commercial ECG features can detect left ventricular hypertrophy with high diagnostic accuracy (AUC up to 0.9859) while offering clinically interpretable feature attributions. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Qibao Zhou; Xiao LUO; Kaihui Du

doi:10.3389/fcvm.2026.1825829

June 4, 2026Frontiers in Cardiovascular MedicineOpen Access

Interpretable detection of left ventricular hypertrophy using commercial ECG features and machine learning: a study based on the PTB-XL+ dataset

View Full Paper

Ask AI

Bookmark

Share

Key result

XGBoost machine learning using commercial ECG features detects LVH with ~0.99 AUC.

n=11,692

Figures

Why the study?

Can interpretable machine learning using commercial ECG features accurately detect left ventricular hypertrophy?

Key Points

This study aims to enhance the detection of left ventricular hypertrophy using interpretable machine learning models applied to ECG data.
Retrospective analysis of 11,692 ECG records from the PTB-XL+ dataset, including 2,258 with LVH and 9,434 normal cases.
Evaluated three feature sets and four classifiers using five-fold stratified cross-validation.
Utilized SHAP analysis for feature importance interpretation.
XGBoost achieved the highest performance with an AUC of 0.9859 for the GE 12SL feature set.
Commercial ECG features slightly outperformed open-source ECGDeli features.
Key features identified included T-wave changes, R-wave amplitude, and QRS area, with age recognized as an important non-ECG predictor.

Study Design

Type

Observational (n=11,692)

Multicenter

No

Structured PICO

Can interpretable machine learning using commercial ECG features accurately detect left ventricular hypertrophy?

P

Population

11,692 ECG records from the PTB-XL+ dataset (2,258 with left ventricular hypertrophy, 9,434 normal)

I

Intervention

Machine learning classifiers (XGBoost, LightGBM, Random Forest, Logistic Regression) using pre-extracted commercial (GE 12SL, UNIG) and open-source (ECGDeli) ECG features

O

Outcome

Detection of left ventricular hypertrophy (measured by Area Under the Curve [AUC])surrogate

Limitations

Dataset originates from a single institutional source, limiting generalizability to diverse populations.
LVH labels are ECG-based diagnostic annotations rather than echocardiographic confirmations, introducing potential label noise.
Binary classification framework does not address the spectrum of LVH severity or distinguish between concentric and eccentric hypertrophy patterns.
Pre-extracted features may not capture all relevant ECG information available in raw waveforms.
Height and weight data were missing for a substantial proportion of the cohort.
Class imbalance may affect model calibration.
Lack of paired cardiac imaging for direct patient-level comparison.

Implication

Interpretable machine learning models using commercial ECG features can detect left ventricular hypertrophy with high accuracy (AUC >0.98) while providing clinically meaningful feature attributions.

Abstract

Introduction and objectives: Left ventricular hypertrophy (LVH) predicts cardiovascular morbidity and mortality. ECG is the most accessible LVH screening tool, but conventional voltage criteria have low sensitivity. Deep learning improves detection but lacks interpretability. This study aims to evaluate the utility of PTB-XL+ pre-extracted commercial ECG features for interpretable LVH detection. Methods: We retrospectively analyzed 11,692 PTB-XL+ ECG records (2,258 LVH, 9,434 normal). Three feature sets (GE 12SL:782, UNIG:702, ECGDeli:529) and four classifiers (XGBoost, LightGBM, Random Forest, Logistic Regression) were evaluated via five-fold stratified cross-validation. SHAP analysis provided feature importance interpretations. Results: XGBoost performed best (AUC: 12SL = 0.9859, UNIG = 0.9853, ECGDeli = 0.9828). Commercial features marginally outperformed open-source ECGDeli. SHAP identified key features (V5-V6T-wave changes, precordial R-wave amplitude, V1 QRS area) and age as an important non-ECG predictor. Conclusion: Interpretable machine learning (ML) with commercial ECG features achieves excellent LVH detection with clinically meaningful attributions, bridging high performance and interpretability for clinical integration.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Zhou et al. (2026) conducted an observational in Left ventricular hypertrophy (n=11,692). XGBoost with GE 12SL ECG features vs. Other machine learning models and feature sets (UNIG, ECGDeli) was evaluated on Area under the receiver-operating-characteristic curve (AUC-ROC) for LVH detection. XGBoost machine learning using GE 12SL commercial ECG features achieved an AUC of 0.9859 for detecting left ventricular hypertrophy.

synapsesocial.com/papers/6a3c2356d15afadd906fa041 https://doi.org/10.3389/fcvm.2026.1825829

View Full Paper

Ask AI

Bookmark

Share

Interpretable detection of left ventricular hypertrophy using commercial ECG features and machine learning: a study based on the PTB-XL+ dataset

Figures

Key Points

Study Design

Structured PICO

Limitations

Implication

Abstract

Discussion

Cite This Study

Interpretable detection of left ventricular hypertrophy using commercial ECG features and machine learning: a study based on the PTB-XL+ dataset

Figures

Key Points

Study Design

Structured PICO

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

More related papers

More from Frontiers in Cardiovascular Medicine

Keep reading

More related papers

More from Frontiers in Cardiovascular Medicine