Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
November 13, 2025PLoS ONEOpen Access

Two-layer machine learning outperforms pooled cohort equations for on-site CHD prediction with ~79% accuracy.

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Few studies quantitatively estimate CHD risk on-site as an auxiliary diagnosis, and machine learning models require validation using real clinical data before adoption in real clinical settings.

Does a two-layer machine learning model improve the accuracy of on-site coronary heart disease prediction compared to pooled cohort equations in high-risk adults?

Population

20,821 patients with CHD and 9,796 controls from a single hospital

Comparison

Two-layer machine learning model vs pooled cohort equations

Design

Retrospective hospital-based cohort study

Key result

A two-layer machine learning model achieved an on-site coronary heart disease prediction accuracy of 0.79 (95% CI 0.79-0.80), outperforming pooled cohort equations (accuracy 0.59).

Authors

LMLiwen MoHLHua LinCLChengxuan Li

Discussion

Loading...

Member takes

Overview

May enhance on-site CHD prediction in high-risk adults; leaves open prospective validation before clinical adoption.

Study Design

Type

Cohort (n=30,617)

Multicenter

No

Structured PICO

Does a two-layer machine learning model improve the accuracy of on-site coronary heart disease prediction compared to pooled cohort equations in high-risk adults?

P
Population
30,617 high-risk adults (20,821 patients with coronary heart disease and 9,796 controls) from a retrospective hospital-based cohort (The Second Affiliated Hospital of Guangxi Medical University) between 2017 and 2024.
I
Intervention
Two-layer machine learning model (TLML) combining random forest and gradient boosting decision tree for on-site prediction of coronary heart disease using clinical data.
C
Comparator
Pooled cohort equations (PCEs).
O
Outcome
Accuracy, sensitivity, and specificity for on-site coronary heart disease prediction.

Main Result

Absolute Event Rate: 0.79% vs 0.59%

A two-layer machine learning model using routine clinical data significantly improves the accuracy of on-site coronary heart disease prediction compared to traditional pooled cohort equations.

Cite This Study

Mo et al. (2025) conducted a cohort in Coronary heart disease (n=30,617). Two-layer machine learning model (TLML) vs. Pooled cohort equations (PCEs) was evaluated on On-site CHD prediction accuracy (95% CI 0.79-0.80). A two-layer machine learning model achieved an on-site coronary heart disease prediction accuracy of 0.79 (95% CI 0.79-0.80), outperforming pooled cohort equations (accuracy 0.59).

synapsesocial.com/papers/6a0514286c3d07813971bd34https://doi.org/10.1371/journal.pone.0334881
View Full Paper
Ask AI
Bookmark
Share