Dynamic Risk Stratification Models for Cardiovascular Disease Prevention: A Systematic Review of Simulation-Based and Clinical Evidence

The clinical evidence prediction models are very paramount in healthcare for estimating the disease probabilities through various predictors. While some studies have focused on deep learning and machine learning for cardiovascular disease (CVD) prevention, only little or none is known about the dynamic risk stratification models for CVD. Thus, this study seeks to address the gap, using the systematic review approach. The study used secondary data collected from five (5) data through a structured search that involve inclusion and exclusion criteria. The Preferred Item for Systematic Review and Meta-analysis and SPIDER tool were used for in the data collection process. A total of seventeen (17) articles were finally selected for the study. Results showed that the dynamic risk stratification models used are characterized by their capacities to integrate time-varying, longitudinal, or multimodal data. Findings showed that the simulation-based techniques used include model disease progression, treatment effects, and physiological mechanisms. Results showed that clinical data such as cohort studies, external validation datasets, and real-world data were used to validate dynamic risk stratification for CVD prevention. The study established that the predictive performance showed an improvement compared to the traditional risk prediction approaches. Results showed that there are several methodological, ethical, and implementation challenges faced in the application of stratification models for cardiovascular disease prevention. It was concluded that dynamic risk stratification models are better than the traditional approaches, especially those using machine learning and simulation-based techniques.