A Method-Centric and Structured Review of Artificial Intelligence-Based Cuffless Blood Pressure Estimation Using ECG and PPG

Abstract

Blood Pressure (BP) is a vital sign that provides information on the cardiovascular health of humans. Traditionally, a sphygmomanometer has been used to measure BP. For conditions such as hypertension, continuous BP monitoring is necessary; however, it is difficult to capture continuous data using a sphygmomanometer. Due to technological developments, Electrocardiography (ECG) and Photoplethysmography (PPG) are used for non-invasive continuous monitoring of BP. The ECG monitors the electrical activity of the heart, whereas the PPG measures blood volume changes in the microvascular bed of tissues. Research shows that Artificial Intelligence (AI) based models can precisely estimate BP from the ECG and PPG signals, which is an alternative and convenient technique against intrusive techniques. In this paper, AI-based BP estimation techniques using ECG, PPG, and their combination are presented as a systematic review. To select relevant studies, a Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guided methodology was used. The selected studies were analysed through an in-depth analysis process using an approach that includes datasets, signal modalities, methodologies, performance metrics, and challenges. From the review, it can be concluded that, especially in the case of multimodal ECG-PPG fusion, Deep Learning (DL) models perform better than traditional Machine Learning (ML) models. In the review, as the foundational steps for powerful AI-based BP estimation, the preprocessing of signals along with feature engineering is highlighted. This systematic review gives thorough understanding into current trends and gaps, and it can be considered as a valuable source of reference work for those working on cuffless BP estimation techniques.