کنفرانس بین المللی کنترل ، ابزار دقیق و اتوماسیون

صفحه اصلی / نهمین کنفرانس بین المللی کنترل ، ابزار دقیق و اتوماسیون

Blood Pressure Estimation through Photoplethysmography and Machine Learning Models

نویسندگان :

Hanieh Mohammadi¹ Bahram Tarvirdizadeh² Khalil Alipour³ Mohammad Ghamari⁴

1- University of Tehran 2- University of Tehran 3- دانشگاه تهران 4- Kettering University

کلمات کلیدی :

blood pressure،feature extraction،machine learning،photoplethysmograph،regression

چکیده :

Blood pressure (BP) is a critical health factor, the fluctuations of which can have profound implications for an individual's well-being. Traditional methods of measuring BP, such as cuff-based and invasive devices, are not only uncomfortable but also unable to provide continuous monitoring. In response to this need, we propose a cuffless, continuous, and non-invasive BP measurement system utilizing photoplethysmograph (PPG) signals and machine learning (ML) models. PPG, an optical volumetric measurement technique, can detect changes in blood volume within the vascular bed of tissue. In this study, PPG signals obtained from diverse individuals underwent preprocessing and feature extraction. Subsequently, feature selection technique were employed to identify suitable features. These selected features were then used to train and evaluate ML models. Ultimately, we identified optimal regression models for independent estimation of systolic blood pressure (SBP) and diastolic blood pressure (DBP). Our results indicate that the random forest (RF) model, in combination with the SelectFromModel feature selection method, outperformed other models. This model yielded significant outcomes, achieving a root mean square error (RMSE) of 10.06 for SBP and 6.61 for DBP, highlighting its superior performance in BP estimation.