Hybrid AI Models for Continuous Vital Sign Monitoring: A Comparative Study of Deep Learning and Statistical Approaches

Abstract

High-risk pregnancies require continuous physiological monitoring to detect early deviations in maternal or fetal health. Traditional clinical monitoring routines, often based on intermittent hospital visits, may fail to capture rapid or unpredictable complications, resulting in delayed interventions. This paper proposes a comprehensive telemonitoring framework integrating remote sensing technologies, maternal–fetal vital tracking, and machine learning–driven predictive analytics to support early detection and timely clinical decision-making. Through the use of wearable devices, Internet of Medical Things (IoMT) networks, and cloud-based analytics, the proposed system enables continuous assessment of fetal heart rate patterns, maternal blood pressure variations, uterine contraction profiles, and biophysical indicators. A case study involving 210 high-risk pregnant women demonstrates notable improvements in anomaly detection, care coordination, and emergency triage. Data analytics reveal that real-time remote monitoring resulted in a 34% reduction in emergency admissions, a 22% reduction in undetected fetal distress, and a 29% improvement in timely intervention compared with standard care. The findings highlight the clinical value, usability, and reliability of telemonitoring for maternal–fetal medicine, while also acknowledging implementation barriers, digital literacy gaps, and challenges in device interoperability.