Intensive Care Units (ICUs) provide continuous monitoring of severely ill patients who may suffer from numerous health complications that affect morbidity and mortality. For clinicians, interpreting data in real time and making decisions is a challenging task. This study is particularly focused on the prediction and monitoring of Mean Blood Pressure (MBP) as a critical physiological parameter. Our approach included the development of a novel data acquisition system that collects real-world time series data from Dash 4000 monitors connected to patients in intensive care units. The use of real clinical data, cleaned and processed for analysis, distinguishes this work from others, enhancing the practicality of our forecasting models in real-world settings. These data were used to develop two predictive models: one based on Long Short-Term Memory (LSTM) networks and another using Multiple Linear Regression. We applied machine learning and deep learning techniques to predict MBP over various time windows, leveraging the multivariate nature of the collected data. The results demonstrate that the Multiple Linear Regression model outperformed the LSTM model. For example, in the 1-min prediction window, the Multiple Linear Regression model achieved an RMSE of 0.45605 and an R \(^2\) score of 0.99053, demonstrating better performance compared to the LSTM model. This system offers promising potential for improving the real-time prediction and monitoring of mean blood pressure in critically ill patients, with practical implications for intensive care settings.