Multistage Electrocardiographic Profiling for Mental Stress-Induced Myocardial Ischemia in Women: A Prospective, Age-Matched Cohort Study in China

Figures

• 

  Figure 1.. Multimodal cohort characteristics and description of ECG data. A, The construction of study and validation cohorts with multimodal data collection. B, Correlation between the measured and predicted summed difference score (SDS) based on multimodal data in validation cohort (n=53 participants). The model was trained on study cohort (n=119). The solid line is the OLS linear regression while the dotted line is measured SDS=predicted SDS. Clinical factors: OLS linear regression model with sleep quality, onset time, hypertension, syncope, and Canadian Cardiovascular Society as regressors; Psychometric tests: OLS linear regression model with hospital anxiety and depression scale (HADS), positive and negative affect schedule (PANAS), state-trait anxiety inventory (STAI), perceived stress scale (PSS), life event scale (LES), PTSD Checklist-Civilian Version (PCL-C) as regressors. ECG variables: the interpretable ECG variables based on literature review, calculated from the original complete ECG data (see in Table S3); Combined modalities: the combination of the above modalities, including clinical factors, psychometric tests, and ECG variables. SDS in PET/CT scans, regarded as the golden…

• 

  Figure 2.. Intrastage ECG variables analysis. A, The significance of ECG variable differences within stages of rest, stress, and recovery across groups. Bubble size and color correspond to P value magnitude and statistical significance. B, The number of intrastage ECG variables exhibiting significant group differences (P<0.05), relating to A. C through E, The Spearman correlation coefficients between intrastage ECG variables and summed difference score (SDS), alongside psychometric tests scores across rest (B), stress (C), and recovery (D) stages. × indicated statistical significance at P<0.05. F, The number of intrastage ECG variables with significant correlations (P<0.05) to SDS or psychometric tests scores, relating to C through E.

• 

  Figure 3.. Interstage ECG variables analysis. A, The significance of interstage ECG variable differences across groups. The interstage variables were derived by directional subtraction across continuous stages stress (see in Method). Bubble size and color correspond to P value magnitude and statistical significance. B, The number of interstage ECG variables exhibiting significant group differences (P<0.05), relating to A. C through E, The Spearman correlation coefficients between interstage ECG variables and summed difference score (SDS), alongside psychometric tests scores across rest-stress (B), stress-recovery (C), and recovery-rest (D). Stars indicated statistical significance at P<0.05. F, The number of interstage ECG variables with significant correlations (P<0.05) to SDS or psychometric tests scores, associated with B through D.

• 

  Figure 4.. ROC-AUC performance across feature sets in the study cohort (n=119) and test cohort (n=53). Each line represents 1 classifier, with feature sets ordered from single-stage inputs to combined multistage variables augmented by interstage difference features. Complete strands for original data which was not divided into 3 stages. R-S stands for derived variables obtained by directional subtraction across continuous stages of Rest-Stress, S-R for Stress-Recovery, R-R for Recovery–Rest. ROC-AUC indicates Receiver Operating Characteristic – Area Under the Curve.