Diagnostic performance of single-lead electrocardiograms for arterial hypertension diagnosis: a machine learning approach

Background Awareness and early identification of hypertension is among the most crucial components in reducing the burden of cardiovascular disease (CVD). Artificial intelligence-based analysis of electrocardiograms (ECGs) can improve diagnosis and patients’ management. ECG technology is now ubiquitous in smartwatches and other wearables and can record a single-lead ECG which is equivalent to lead I on a standard 12-lead ECG. Our aim was to develop machine learning (ML) algorithms to diagnose hypertension from single-lead ECGs. Methods We enrolled 1254 subjects (539 male, aged 60.22±12.46 years),with and without essential hypertension, and no indications of CVD. A 12-lead ECG of 10 seconds duration in resting position was performed on each subject using a digital electrocardiograph. A Random Forest (RF) was then trained on features derived only from lead I tracings. Results The RF model was able to distinguish hypertensive from normotensive subjects on a hold-out test set, with 81% accuracy, AUC/ROC 0.82, sensitivity 88%, and specificity 70% (both using a threshold of 0.5) using only features from lead I. Age, body mass index, the area under the T wave divided by the QRS complex area, the area under QRS segment adjusted for BMI, the QT corrected interval, and the PQ interval, were the most important anthropometric and ECG-derived features in terms of the success of our model (Figure 1). Conclusions This study demonstrates the promising potential of detecting an undiagnosed hypertension opportunistically, using a single-lead ECG. Our findings offer the opportunity to evolve innovative technologies to improve hypertension awareness at the community level.