This study investigates the relationship between photoplethysmography (PPG) signals and central blood pressure (CBP) using a novel cardiovascular simulator with a custom wrist phantom, examining how heart rate, stroke volume, and peripheral resistance affect the accuracy of system identification models in predicting CBP from PPG signals. Results demonstrate that ensemble averaging performs well for stroke volume and peripheral resistance, with CBP waveform prediction accuracy of 93.06% and 95.38% respectively. However, ensemble averaging had poor performance for heart rate variations with an accuracy of only 83.65%. This suggests that parameter-specific modeling approaches may be necessary for developing accurate, non-invasive CBP monitoring systems using wearable PPG.Clinical relevanceUnderstanding how cardiovascular parameters influence PPG-to-CBP signal conversion can improve the accuracy of wearable PPG devices for continuous central blood pressure monitoring.