A Comprehensive Review of Machine Learning Techniques for Non-Invasive Blood Pressure Prediction and Hypertension Detection

This survey highlights the significant progress achieved in blood pressure prediction and hypertension detection through machine learning and deep learning techniques. The reviewed studies demonstrate that data-driven models outperform traditional methods by effectively capturing complex and nonlinear relationships among physiological, clinical, and demographic features. Non-invasive sensing approaches, particularly those based on photoplethysmography signals, combined with advanced deep learning architectures, have shown strong potential for continuous and real-time monitoring. The adoption of explainable artificial intelligence methods further improves model transparency and clinical trust by identifying key risk factors influencing predictions. However, challenges such as data imbalance, limited external validation, and variability in datasets remain unresolved. Future research should focus on developing standardized evaluation protocols, improving model generalizability, and validating these systems across diverse populations and clinical environments. Overall, machine learning–based frameworks represent a promising and effective direction for early hypertension detection, personalized healthcare, and improved cardiovascular disease management.

[1] Muhammad Areeb et. al. “An ML Framework for the Early Detection and Prediction of Hypertension: Enhancing Diagnostic Accuracy”  Eng. Proc. 2025, 107, 18 https://doi.org/10.3390/engproc2025107018, MPDI 2025   [2] Devangam Sai Chaithanya1, et al “Hybrid Machine Learning Model for Hypertension Detection” ISSN (O) 2278-1021, IJARCCE ISSN (P) 2319-5940 2025   [3] Mohammed Attya “Performance Evaluation of Deep Learning Architectures for Blood Pressure Estimation Using Photoplethysmography” I.J. Intelligent Systems and Applications, 2024, 5, 22-38 DOI: 10.5815/ijisa.2024.05.03   [4] Edo Septian1 et al “Prediction of Personalised Hypertension Using Machine Learning in Indonesian Population” Journal of Medical Systems (2025) https://doi.org/10.1007/s10916-025-02253-5   [5] Zheng Zhang et al “Machine learning prediction and explanation of high intraoperative blood pressure variability for noncardiac surgery using preoperative factors” Zhang et al. BMC Cardiovascular Disorders (2025)  https://doi.org/10.1186/s12872-025-05026-7   [6] Ekaba Bisong1, “Predicting high blood pressure using machine learning models in lowand middle‑income countries”  BMC Medical Informatics and Decision Making (2024)  https://doi.org/10.1186/s12911-024-02634-9   [7] Qianyu Liu1 et. al. “Photoplethysmography‑based non‑invasive blood pressure monitoring via ensemble model and imbalanced dataset processing” Physical and Engineering Sciences in Medicine (2024) 47:1307–1321 https://doi.org/10.1007/s13246-024-01445-6   [8] Stevens SL, Wood S, Koshiaris C, Law K, Glasziou P, Stevens RJ, et al. Blood pressure variability and cardiovascular disease: systematic review and metaanalysis. BMJ. 2016;354:i4098.   [9] Hammarstrom S. The variability of blood pressure in hypertension. Acta Med Scand. 1948;131:94–101.   [10] Putowski Z, Czok M, Krzych ŁJ. The impact of intraoperative blood pressure variability on the risk of postoperative adverse outcomes in non-cardiac surgery: a systematic review. J Anesth. 2022;36:316–22.