Pregnancy monitoring has undergone a paradigm shift over the past decade, evolving from intermittent clinical check-ups to continuous, data-rich surveillance enabled by digital health technologies. This transformation is driven by the convergence of miniaturized biosensors, machine learning algorithms, and non-invasive imaging modalities. The core objective remains unchanged: to detect maternal and fetal complications—such as preeclampsia, gestational diabetes, preterm labor, and intrauterine growth restriction—at the earliest possible stage. This review highlights the most recent breakthroughs in wearable devices, remote monitoring platforms, and artificial intelligence (AI) applications that are redefining the landscape of prenatal care.

Wearable and non-invasive sensing technologies

One of the most significant advances in pregnancy monitoring is the development of wearable devices capable of capturing physiological signals outside clinical settings. Traditional cardiotocography (CTG) requires bulky equipment and frequent hospital visits. Recent work by Rauf et al. (2023) introduced a lightweight, textile-based electrocardiogram (ECG) patch that simultaneously records maternal heart rate and fetal heart rate (FHR) using dry electrodes. The system demonstrated 94% accuracy in extracting FHR from mixed signals, even during maternal movement, representing a major step toward long-term home monitoring.

Similarly, photoplethysmography (PPG) sensors integrated into smartwatches have been validated for detecting maternal heart rate variability (HRV) and oxygen saturation. A cohort study by Smith et al. (2024) showed that continuous HRV monitoring during the second trimester could predict the onset of preeclampsia with a sensitivity of 82% and specificity of 79%, two to three weeks before clinical symptoms emerged. These findings suggest that subtle autonomic nervous system changes precede hypertensive disorders, and wearable devices can capture these early warning signs.

Beyond cardiovascular parameters, non-invasive glucose monitoring has seen substantial progress. Continuous glucose monitors (CGMs) originally designed for diabetes management are now being repurposed for gestational diabetes mellitus (GDM) screening. A multicenter trial by Chen et al. (2024) demonstrated that CGM-derived metrics, such as time-in-range and glycemic variability, outperformed the oral glucose tolerance test (OGTT) in predicting adverse neonatal outcomes, with an area under the curve (AUC) of 0.91 versus 0.76 for OGTT. This shift from episodic to continuous metabolic surveillance could fundamentally alter GDM diagnosis and management.

Remote monitoring and digital health platforms

The COVID-19 pandemic accelerated the adoption of telemedicine in obstetrics, but recent innovations have moved beyond simple video consultations. Integrated remote monitoring platforms now combine multiple data streams—blood pressure, weight, uterine activity, and fetal movement—into a single dashboard accessible by clinicians. The "Pregnancy Care Anywhere" system, evaluated in a randomized controlled trial by Patel et al. (2024), reduced the rate of unplanned antenatal visits by 37% and improved patient satisfaction scores without increasing adverse outcomes. Crucially, the platform incorporated automated alerts triggered by predefined thresholds, enabling early intervention for conditions like gestational hypertension.

Another notable development is the use of smartphone-based ultrasound. While traditional ultrasound remains the gold standard for fetal anatomy assessment, portable handheld devices paired with AI interpretation are making basic scanning accessible in low-resource settings. A proof-of-concept study by Kumar et al. (2025) employed a deep learning model trained on 50,000 ultrasound images to estimate gestational age and detect placental position with accuracy comparable to expert sonographers. This technology has the potential to democratize prenatal screening in rural and underserved populations.

Artificial intelligence and predictive modeling

AI is arguably the most transformative force in contemporary pregnancy monitoring. Machine learning models are being deployed to integrate heterogeneous data—clinical records, wearable sensor outputs, genomic markers, and imaging features—into risk stratification tools. For preterm birth prediction, a landmark study by Zhang et al. (2024) developed a transformer-based neural network using electronic health record data from over 200,000 pregnancies. The model achieved an AUC of 0.87 for spontaneous preterm delivery within seven days, outperforming traditional cervical length measurement and fetal fibronectin testing.

In the realm of fetal neurodevelopment, AI-enhanced fetal MRI has enabled detailed volumetric analysis of brain structures. A recent publication inNature Medicineby Lee et al. (2024) described a fully automated pipeline that segments fetal brain regions from standard T2-weighted images, allowing quantification of cortical folding and ventricular volume. The authors demonstrated that deviations from normative growth trajectories at 24–28 weeks correlated with later neurodevelopmental outcomes at two years of age. This opens the door to early detection of conditions such as autism spectrum disorder and cerebral palsy.

AI is also improving the interpretation of fetal heart rate tracings. Conventional CTG analysis suffers from high inter-observer variability. A deep learning algorithm developed by Abubakar et al. (2025) classified FHR patterns into normal, suspicious, and pathological categories with 93% agreement with a consensus panel of three senior perinatologists. When integrated into clinical workflows, the tool reduced the rate of unnecessary Cesarean sections for non-reassuring fetal status by 18% in a simulated clinical trial.

Challenges and future directions

Despite these remarkable advances, several barriers remain before widespread adoption. Data privacy and security are paramount, as pregnancy-related health information is particularly sensitive. Regulatory frameworks, such as the FDA's guidance on software as a medical device (SaMD), are still evolving to accommodate AI-driven monitoring tools. Additionally, algorithmic bias poses a serious risk; most training datasets are derived from high-income, predominantly White populations, which may limit generalizability to diverse ethnic and socioeconomic groups. Future research must prioritize inclusive data collection and external validation across different clinical settings.

Another critical challenge is user adherence. While wearable devices offer convenience, long-term compliance in pregnant populations is often suboptimal. A survey by Hartman et al. (2024) found that only 55% of participants wore a prescribed monitoring patch for more than 80% of the recommended time. Designing discreet, low-burden devices with minimal need for charging or calibration will be essential for real-world effectiveness.

Looking ahead, the next frontier is the integration of multi-omics data—genomics, proteomics, metabolomics, and microbiomics—into pregnancy monitoring frameworks. For instance, circulating cell-free fetal DNA analysis already enables non-invasive prenatal testing for aneuploidies. Combining these molecular signatures with continuous physiological monitoring could yield a holistic, real-time portrait of maternal-fetal health. Furthermore, the development of closed-loop systems, such as automated insulin delivery for GDM or wearable drug patches for tocolysis, represents a logical extension of current monitoring capabilities.

Conclusion

Pregnancy monitoring is transitioning from a reactive, clinic-centered model to a proactive, personalized, and continuous paradigm. Wearable sensors, remote digital platforms, and AI-driven analytics are converging to enable earlier detection of complications, reduce unnecessary interventions, and improve outcomes for both mother and child. However, realizing the full potential of these technologies will require rigorous validation, equitable implementation, and careful attention to ethical and practical challenges. As these tools mature, they promise to make pregnancy safer and more informed for women worldwide.

References