Detection of Acute Myocardial Infarction Using Deep Learning on Lead-I ECG Data

Q: What is the clinical evidence from this study?

Study design: Observational. Population: Myocardial Infarction (n=6586). Intervention: Deep learning model (ResNet16) on Lead-I ECG data vs. 12-lead ECG configuration. Primary outcome: Area under the receiver-operating curve (AUC) for MI detection.

Parmida Davarmanesh; Qian Lin; Irene Tenison; Gabriel Jabbour; Ridwan Alam

doi:10.1101/2024.10.15.24315544

October 16, 2024Open Access

Detection of Acute Myocardial Infarction Using Deep Learning on Lead-I ECG Data

View Full Paper

Ask AI

Bookmark

Share

Key result

Lead-I ECG deep learning detects MI with ~0.92 AUC, slightly trailing standard 12-lead configurations.

n=6,586

Why the study?

Early detection of MI is crucial for timely intervention, prompting investigation into whether deep learning on lead-I ECG data can feasibly predict MI for wearable monitoring systems.

Does a deep learning model using only lead-I ECG data accurately detect acute myocardial infarction compared to multi-lead ECG configurations?

Key Points

To assess the feasibility of detecting acute myocardial infarction using single lead-I ECG data via deep learning to support wearable cardiac monitoring applications.
Extracted 12-lead ECG records including normal and diverse myocardial infarction cases from the PTB-XL dataset.
Trained and evaluated deep learning binary classification models using single-lead (lead-I), 6-lead, and 12-lead configurations.
Single lead-I ECG models achieved an area under the receiver operating characteristic curve (AUC) of 0.92 and an area under the precision-recall curve (AUPR) of 0.82 on the test set.
Multi-lead setups using 6-lead and 12-lead ECG data both achieved an AUC of 0.99.

Study Design

Type

Observational (n=6,586)

Structured PICO

Does a deep learning model using only lead-I ECG data accurately detect acute myocardial infarction compared to multi-lead ECG configurations?

P

Population

6,586 10-second clinical ECG records from the PTB-XL dataset, including 5,848 normal and 738 myocardial infarction (MI) samples. Inclusion required 100% diagnostic confidence validated by at least one cardiologist.

I

Intervention

Deep learning binary classification model (1-D ResNet16) trained and evaluated using only lead-I ECG data.

C

Comparator

Deep learning models trained on 6-lead (I, II, III, aVF, aVL, aVR), 12-lead, or other single-lead ECG configurations.

O

Outcome

Detection of Myocardial Infarction (measured by Area Under the Receiver-Operating Curve [AUC], Area Under the Precision-Recall Curve [AUPR], sensitivity, specificity, and accuracy).

Main Result

Absolute Event Rate: 0.92% vs 0.99%

Deep learning applied to single-lead (lead-I) ECG data can detect acute myocardial infarction with high accuracy (AUC 0.95), demonstrating the potential for wearable devices to enable early MI detection.

Limitations

Significant imbalance of outcomes in the training data (6.8% MI)
Lack of external validation on other datasets
Potential performance degradation when applied to inherently noisier ECG signals acquired directly from wearable devices
Lack of validation on actual wearable device ECG signals, which inherently suffer from higher noise levels

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

Supports wearable MI detection via lead-I ECG; extends deep learning applications beyond standard 12-lead configurations.

Abstract

Myocardial Infarction (MI) is a major global health concern due to its high mortality and morbidity rates. Early detection of MI is crucial for timely medical intervention and improved patient outcomes. In this study, we investigate the feasibility of predicting MI using lead-I of electrocardiogram (ECG) data, with a focus on practical applications for wearable monitoring systems. Utilizing the PTB-XL dataset, which includes a comprehensive collection of 12-lead ECG recordings with both normal and various MI samples, we employ deep learning techniques to develop a binary classification model. For MI detection using lead-I, we achieved an AUC of 0.92 and an AUPR of 0.82 on the test set. In comparison, using 6-lead and 12-lead configurations both resulted in an AUC of 0.99. These findings demonstrate the potential for detecting MI using only lead-I, as measured by wearable devices. This advancement could significantly enhance clinical outcomes for MI patients by enabling timely detection and intervention.

Discussion

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

Cite This Study

Davarmanesh et al. (2024) conducted an observational in Myocardial Infarction (n=6,586). Deep learning model (ResNet16) on Lead-I ECG data vs. 12-lead ECG configuration was evaluated on Area under the receiver-operating curve (AUC) for MI detection. A deep learning model utilizing only lead-I ECG data detected myocardial infarction with an AUC of 0.92, compared to an AUC of 0.99 using standard 12-lead configurations.

synapsesocial.com/papers/6a089c359a6c4ba6e610bdd8 https://doi.org/10.1101/2024.10.15.24315544

View Full Paper

Ask AI

Bookmark

Share

Detection of Acute Myocardial Infarction Using Deep Learning on Lead-I ECG Data

Key Points

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Detection of Acute Myocardial Infarction Using Deep Learning on Lead-I ECG Data

Key Points

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading