Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
July 8, 2026Frontiers in Cardiovascular MedicineOpen Access

Development and validation of an interpretable machine learning model for predicting atrial fibrillation risk in middle-aged and older patients with coronary heart disease

View Full Paper
Ask AI
Bookmark
Share

Key result

An interpretable XGBoost machine learning model accurately predicted the risk of prevalent atrial fibrillation in middle-aged and older patients with coronary heart disease, achieving an AUC of 0.813 in the validation set.

Why the study?

Existing risk stratification tools inadequately capture the nonlinear, multidimensional determinants of AF in middle-aged and older CHD patients.

Does an interpretable machine learning model (XGBoost) accurately predict prevalent atrial fibrillation in middle-aged and older hospitalized patients with coronary heart disease?

Population

47,617 hospitalized CHD patients

Comparison

Eight machine learning algorithms for AF risk prediction

Design

Retrospective cohort study

Authors

FCFeng ChenQFQin FuLLLi Li

Discussion

Loading...

Member takes

Overview

May aid AF risk stratification in CHD; hypothesis-generating and requires prospective validation before clinical use.

Study Design

Type

Cross-Sectional (n=47,617)

Multicenter

No

Structured PICO

Does an interpretable machine learning model (XGBoost) accurately predict prevalent atrial fibrillation in middle-aged and older hospitalized patients with coronary heart disease?

P
Population
47,617 hospitalized middle-aged and older adults (median age 74, 48.3% female) with coronary heart disease, evaluated for prevalent atrial fibrillation during their index hospitalization.
E
Exposure
Interpretable machine learning-based prediction model (XGBoost) leveraging electronic medical records (EMR) data
C
Comparator
Seven other machine learning algorithms (GBDT, LightGBM, RF, AdaBoost, LR, DT, and NB)
O
Outcome
Presence of atrial fibrillation (AF) during the index hospitalization, defined by ICD-10 criteria and confirmed by at least one 12-lead electrocardiogram (ECG) report or continuous ECG monitoring record

Main Result

Effect estimate: AUC 0.813 (95% CI 0.802-0.823)

p-value: p=<0.001

An interpretable XGBoost machine learning model using routine EMR data accurately identifies prevalent atrial fibrillation in middle-aged and older patients hospitalized with coronary heart disease.

Limitations

  • Retrospective, single-center design limits generalizability
  • Cross-sectional design precludes causal inference and evaluates prevalent rather than incident AF
  • Lack of longitudinal follow-up
  • Potential for unmeasured confounders despite comprehensive EMR data extraction

Cite This Study

Chen et al. (2026) conducted a cross-sectional in Coronary heart disease (CHD) (n=47,617). XGBoost machine learning model vs. Other machine learning models (GBDT, LightGBM, RF, AdaBoost, LR, DT, NB) was evaluated on Presence of atrial fibrillation (AF) during the index hospitalization (AUC 0.813, 95% CI 0.802-0.823, p=<0.001). An interpretable XGBoost machine learning model accurately predicted the risk of prevalent atrial fibrillation in middle-aged and older patients with coronary heart disease, achieving an AUC of 0.813 in the validation set.

synapsesocial.com/papers/6a89a64b5edafe343d92ee13https://doi.org/10.3389/fcvm.2026.1886992
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Clinical, Electrocardiographic and Echocardiographic Predictors of Atrial Fibrillation Recurrence After Pulmonary Vein Isolation2025 · 8 citations
  2. 2Association of Wearable Device Use With Pulse Rate and Health Care Use in Adults With Atrial Fibrillation2021 · 46 citations
  3. 3The association between the cystatin C- and creatinine-based estimated GFR ratio and post-ablation outcomes in patients with atrial fibrillation2025 · 1 citations
  4. 4Inflammatory and fibrotic signaling pathways mediated by cardiac macrophages in atrial fibrillation2026 · 5 citations
  5. 5MIXED-EFFECTS XGBOOST WITH GROUP-AWARE PERMUTATION IMPORTANCE AND CROSS-VALIDATION FOR MULTILEVEL CROSS-CLASSIFIED CONTINUOUS OUTCOMES2026 · 1 citations