Artificial intelligence technology applied to an unoccluded catheter for recognition of hemodynamic changes

This study seeks to provide a reliable means for physicians to diagnose hemorrhagic complications associated with patent ductus arteriosus (PDA) in extremely low birth weight (ELBW) infants. We employ the YOLOv4 algorithm to analyze cardiac parameters from the echocardiography device's ultrasonic wave, such as left ventricular ejection time (LVET), left ventricular internal dimension-diastole (LVIDd), left ventricular internal dimension-systole (LVIDs), posterior wall thickness at end-systole (HES), and RR interval. The proposed model has achieved an impressive average detection accuracy of 89.4% for these five parameters. Additionally, the further evaluation of cardiac function can be done by deriving the end-systolic wall stress (ESWS) and left ventricle rate-corrected mean velocity of circumferential fiber shortening (mVcFc) values from these parameters. As a result, it indicates that the accuracy of ESWS and mVcFc values is 88.7% and 90.3%, respectively. With this deep learning-based approach, physicians can confidently evaluate the cardiac function of ELBW infants and diagnose PDA-related hemorrhagic complications.