| Abstract [eng] |
A wrist-worn electrocardiogram (ECG) recording device capable of acquiring both limb and precordial ECG leads through a simple touch interaction represents a promising approach for acute myocardial infarction (AMI) detection outside clinical settings. This study investigated whether incorporating peripheral oxygen saturation (SpO₂) together with ECG features could improve AMI detection results. ECG signals were acquired using a wrist-worn device equipped with three electrodes. Three ECG lead configurations were formed: I & V3–LA, I & V5–LA, and I & A–LA. The study dataset consisted of 111 subjects divided into three groups: AMI patients, patients with other cardiovascular diseases, and healthy controls. SpO₂ values were acquired using a pulse oximeter simultaneously with ECG signals within 24 hours after percutaneous coronary intervention. Gradient-boosted decision tree models were implemented for AMI detection. For each lead configuration, two feature sets were evaluated: ECG features alone and ECG features combined with SpO₂. The inclusion of SpO₂ increased the AUC in all lead configurations: from 0.72 to 0.78 for I & V3–LA, from 0.82 to 0.84 for I & V5–LA, and from 0.80 to 0.85 for I & A–LA. The findings suggest that SpO₂ can provide complementary information for AMI detection when used together with ECG features. However, its added value was smaller when distinguishing AMI patients from patients with other cardiovascular diseases. Further research should evaluate the added value of SpO₂ in larger study populations and under conditions where this parameter is acquired by the same wrist-worn device as the ECG signals. |
