Abstract
Smoking has been widely identified for its detrimental effects on human health, particularly on the cardiovascular health. The prediction of these effects can be anticipated by monitoring the dynamic changes in vital signs and other physiological signals or parameters such as heart rate, blood pressure (BP), Electrocardiogram (ECG), and Photoplethysmogram (PPG), which subtly encode smoking-related effects. We investigated the influence of different smoking habits-normal cigarettes (NC), electronic cigarettes (EC), and shisha (SH)-on BP through analysis of ECG and PPG signals. The measurements of these physiological signals were taken across three distinct smoking phases: "before", "during", and "after" smoking. The study assessed changes in heart rate, as well as morphological and statistical characteristics of ECG and PPG signals, induced by smoking. A machine learning (ML) model was developed to predict BP values with different smoking habits and smoking phases, while also evaluating the temporal effects of smoking phases. Results show that smoking markedly alters PPG features in such it significantly affects systolic time, heart rate, peak pulse interval variability, and augmentation index. BP variations were evident across all smoking habits and phases. The ML model demonstrated strong accuracy in estimating systolic blood pressure (SBP) and diastolic blood pressure (DBP) during and post-smoking, with a mean error of 0.01 ± 0.29 mmHg and a root mean square error (RMSE) of 0.2924 mmHg for DBP and RMSE of 0.0082 mmHg for SBP. Such a study underscores the pronounced effect of smoking on BP and its potential role in cardiovascular system alterations, offering insights into the development of related diseases.