ABSTRACT
Aims: Incorporation of sex in study design can lead to discoveries in medical research. Deep neural networks (DNNs) accurately predict sex based on the electrocardiogram (ECG) and we hypothesized that misclassification of sex is an important predictor for mortality. Therefore, we first developed and validated a DNN that classified sex based on the ECG and investigated the outcome. Second, we studied ECG drivers of DNN-classified sex and mortality. Methods and results: A DNN was trained to classify sex based on 131 673 normal ECGs. The algorithm was validated on internal (68 500 ECGs) and external data sets (3303 and 4457 ECGs). The survival of sex (mis)classified groups was investigated using time-to-event analysis and sex-stratified mediation analysis of ECG features. The DNN successfully distinguished female from male ECGs {internal validation: area under the curve (AUC) 0.96 [95% confidence interval (CI): 0.96, 0.97]; external validations: AUC 0.89 (95% CI: 0.88, 0.90), 0.94 (95% CI: 0.93, 0.94)}. Sex-misclassified individuals (11%) had a 1.4 times higher mortality risk compared with correctly classified peers. The ventricular rate was the strongest mediating ECG variable (41%, 95% CI: 31%, 56%) in males, while the maximum amplitude of the ST segment was strongest in females (18%, 95% CI: 11%, 39%). Short QRS duration was associated with higher mortality risk. Conclusion: Deep neural networks accurately classify sex based on ECGs. While the proportion of ECG-based sex misclassifications is low, it is an interesting biomarker. Investigation of the causal pathway between misclassification and mortality uncovered new ECG features that might be associated with mortality. Increased emphasis on sex as a biological variable in artificial intelligence is warranted.
ABSTRACT
BACKGROUND: Patients with chronic obstructive pulmonary disease (COPD) have an independent increased risk of cardiovascular (CV) disease. Cardiovascular risk (CVR) assessment should be offered to all patients with COPD, according to the new Dutch CVR management (CVRM) guideline (May 2019). AIM: To evaluate the impact of the new CVRM guideline on the care of patients with COPD in primary care. DESIGN & SETTING: A retrospective study took place within five primary healthcare centres located in The Netherlands. METHOD: In accordance with the guideline, the CVR of all patients with COPD was estimated and categorised. Data from 2014-2019 were used for the qualitative risk assessment based on comorbidities, and the quantitative Systematic Coronary Risk Evaluation (SCORE). In addition, the guideline-based follow-up was investigated. RESULTS: Of the 391 patients with COPD, 84.1% (n = 329) had complete data on CVR assessment: 90.3% (n = 297) had a (very) high risk, and 9.7% (n = 32) a low-to-moderate risk. Of the patients with (very) high risk, 73.4% (n = 218) received guideline-based follow-up (primary care: 95.4%, secondary care: 4.6%). In 15.9% (n = 62) of all patients with COPD, the CVR profile was not measured and of the (very) high-risk patients, 26.6% (n = 79) were not enroled in a CV care programme. CONCLUSION: Whereas in the majority of patients with COPD the CVR is already known, for one out of six patients this CVR still has to be assessed according to the recently updated guideline. Moreover, once a (very) high risk has been assessed, as a consequence CV treatment of risk factors should be intensified in one out of four patients with COPD. Adherence to the new CVRM guideline could provide improvement in CVRM in more than a third of all patients with COPD.
