August 3, 2023Scientific ReportsOpen Access

Machine learning approaches that use clinical, laboratory, and electrocardiogram data enhance the prediction of obstructive coronary artery disease

Q: What are the key findings of this study?

An ensemble machine learning model incorporating clinical, laboratory, and electrocardiogram data achieved an AUC of 0.767 for predicting obstructive coronary artery disease, outperforming traditional pretest probability models.

Q: What is the clinical evidence from this study?

Study design: Observational. Population: Suspected obstructive coronary artery disease (n=7907). Intervention: Ensemble machine learning model (clinical, laboratory, and ECG data) vs. Traditional pretest probability models (CAD1, CAD2, PCE) and individual clinical/ECG models. Primary outcome: Area under the receiver operating characteristic curve (AUC) for predicting obstructive coronary artery disease (AUC 0.767, 95% CI 0.758-0.776, p=<0.05).

View Full Paper

Ask AI

Bookmark

Key result

Ensemble machine learning outperforms traditional pretest probability in predicting obstructive CAD, achieving ~0.77 AUC.

AUC 0.767
95% CI 0.758-0.776
P<0.05
n=7,907

Why the study?

Standard laboratory findings and raw ECG data have not been evaluated for integration into pretest probability estimation to reduce ObCAD overestimation.

Does an ensemble machine learning model using clinical, laboratory, and ECG data improve the prediction of obstructive coronary artery disease in patients with suspected ObCAD?

Population

7907 patients with suspected ObCAD who underwent coronary angiography

Comparison

Ensemble ML and DL model vs standalone clinical-laboratory, ECG, and traditional PTP models

Design

Retrospective electronic medical records cohort study

Authors

HLHyun-Gyu LeeInha University SPSang‐Don ParkInterventional / Structural Cardiology JBJang‐Whan BaeInterventional Cardiology

Discussion

Member takes

Implication

May improve CAD pretest probability estimation; leaves open prospective validation before clinical adoption.

Study Design

Type

Observational (n=7,907)

Multicenter

Structured PICO

Does an ensemble machine learning model using clinical, laboratory, and ECG data improve the prediction of obstructive coronary artery disease in patients with suspected ObCAD?

Population

7,907 adult patients with suspected obstructive coronary artery disease who underwent coronary angiography at a single center in South Korea.

Exposure

Ensemble machine learning model combining clinical, laboratory (27 variables), and ECG waveform data.

Comparator

Traditional pretest probability (PTP) models, as well as standalone clinical-laboratory and ECG models.

Outcome

Prediction of obstructive coronary artery disease (ObCAD) measured by Area Under the Curve (AUC).surrogate

Main Result

Effect estimate: AUC 0.767 (95% CI 0.758-0.776)

Absolute Event Rate: 0.767% vs 0.693%

p-value: p=<0.05

An ensemble machine learning model integrating clinical, laboratory, and ECG data improves the prediction of obstructive coronary artery disease, potentially reducing pretest probability overestimation.

Limitations

The performance of the ensemble model was modest and requires further enhancement for practical use.
The study included only patients who underwent coronary angiography, not all patients with suspected ObCAD, which may introduce selection bias.
The electronic medical records could not differentiate between stable and unstable angina, so unstable angina was not excluded.

Cite This Study

Lee et al. (2023) conducted an observational in Suspected obstructive coronary artery disease (n=7,907). Ensemble machine learning model (clinical, laboratory, and ECG data) vs. Traditional pretest probability models (CAD1, CAD2, PCE) and individual clinical/ECG models was evaluated on Area under the receiver operating characteristic curve (AUC) for predicting obstructive coronary artery disease (AUC 0.767, 95% CI 0.758-0.776, p=<0.05). An ensemble machine learning model incorporating clinical, laboratory, and electrocardiogram data achieved an AUC of 0.767 for predicting obstructive coronary artery disease, outperforming traditional pretest probability models.

synapsesocial.com/papers/6a5f2cd78fc5ccacc97e3204 https://doi.org/10.1038/s41598-023-39911-y

Topics

Coronary artery disease

View Full Paper

Ask AI

Bookmark

Machine learning approaches that use clinical, laboratory, and electrocardiogram data enhance the prediction of obstructive coronary artery disease

Discussion

Member takes

Implication

Study Design

Structured PICO

Main Result

Limitations

Cite This Study

Topics

Machine learning approaches that use clinical, laboratory, and electrocardiogram data enhance the prediction of obstructive coronary artery disease

Discussion

Member takes

Implication

Study Design

Structured PICO

Main Result

Limitations

Cite This Study

Also Consider

Topics

Also Consider