Artificial Womb Research Gets a Lifelike Fetal Manikin Wired to a Digital Twin

Engineers in the Netherlands have built a hybrid fetal manikin designed to bridge the gap between animal experiments and clinical translation of artificial womb technology. The device, described in BMC Biomedical Engineering, is a silicone-skinned model of an extremely premature infant whose embedded sensors and actuators are connected over a wireless link to a digital twin of fetal cardiovascular physiology. The team, led by researchers affiliated with Eindhoven University of Technology and Máxima Medical Center, reports that the proof-of-concept simulator can display breathing movements, cyanosis, limb motion, changes in cerebral oxygenation and umbilical blood flow, while monitoring temperature, airway aeration and circulation in real time.

The clinical motivation is stark. Every year roughly 15 million babies are born prematurely worldwide, and about 5 percent of them arrive extremely premature, between 24 and 28 weeks of gestation. Despite intensive care including mechanical ventilation and gastrointestinal nutrition, mortality in this group runs at 30 to 50 percent, and among survivors 20 to 50 percent experience morbidity. Beyond short-term complications such as respiratory distress syndrome, bronchopulmonary dysplasia, necrotizing enterocolitis and intraventricular hemorrhage, extremely preterm birth can leave lasting neurocognitive developmental disorders. Extra-uterine life support technology, in which a perinate is transferred into a liquid-filled artificial womb environment with an artificial placenta, aims to keep underdeveloped lungs maturing in their fetal state while avoiding the damage caused by air-based ventilation.

Preclinical animal studies have demonstrated that fetuses can be maintained on such systems, but the authors argue that the step to human translation remains immense. Lambs at a developmental stage comparable to extremely premature infants are notably larger than human fetuses and have two umbilical arteries and two veins, whereas pigs share the three-vessel cord of human fetuses but have distinctly different placental structures. More human-like organisms such as primates would give better predictions but require extensive pilot studies before ethics committees would approve their use. Future in-human studies would also face unpredictable preterm births, high therapeutic uncertainty and a heavy emotional load on families and staff. A manikin, the researchers reasoned, could offer a low-risk environment for developing the technology, refining procedures and training clinicians before any patient is involved.

The anatomy of the manikin was derived from high-resolution MRI scans of a growth-retarded fetus at 24 weeks of gestation. The segmented three-dimensional model was scaled by a factor of 1.25 to represent an average-sized 24-week fetus with a crown-to-heel length of 29.3 centimeters, in line with published biometry. Segmentation was performed with semi-automatic, pixel-density-based processing in Mimics software, and the skin surface was refined in 3-Matic and 3ds Max, with the face reconstructed to preserve anonymity. A two-part outer mold was 3D printed in Vero Clear on an Objet Connex 350 printer, and a silicone skinsuit was cast in an inner mold using Eurosil 13 pigmented with SilTone skin tones, then airbrushed to achieve realistic coloring, including fine details such as surface veins.

Internal structures were built with the same MRI-to-geometry pipeline. The upper airways, including the oropharynx, larynx, trachea and lungs, were segmented, 3D printed and suspended in the mold, with the lung reconstructed as a single cavity matching the volume of a 24-week fetus; this allows simulations in which the lungs are kept in a fetal state via a liquid-filled endotracheal tube. The skull was printed in a material verified to mimic tissue optics so that near-infrared spectroscopy probes can be tested against it. A flexible thermoplastic polyurethane ribcage was chosen after earlier prototypes with solid sterna and flexible ribs proved too fragile. Artificial umbilical arteries were produced by layering silicone cylindrically on a custom spinning machine, and the 5-millimeter vein was cast from Ecoflex Gel, with the components wound together around a Wharton’s jelly mimic into a simplified single-artery cord. The complete manikin occupies 540 cubic centimeters, of which the head accounts for 221 and the torso 230, and the skinsuit with hardware weighs 350 grams, well below the 665-gram average for a real 24-week fetus.

Because space inside the tiny body is scarce, the team compared actuation options including fluidic and electromagnetic mechanisms before selecting electromechanical servos for cyanosis, limb movement and the breathing reflex. Breathing is driven by a linear servo with an 8-millimeter stroke pulling on the sternum; bench measurements of inspiratory and expiratory velocities and distances averaged 5 millimeters per second and 3 millimeters, within published biometric ranges for mid-gestation fetuses. Cyanosis is displayed by a rotating cone, shaped like a spinning top, painted with a gradient from non-cyanotic to cyanotic tones based on patient color data, visible through a transparent window in the skull and skin. The cerebral saturation phantom places rotating black PVC rods of three diameters under an NIRS probe to simulate falling oxygenation, and its optical properties were characterized by time-of-flight measurements, yielding absorption coefficients of 0.11 and 0.13 per centimeter at 685 and 830 nanometers respectively. Limb motion is produced by four compliant thermoplastic flexures allowing flexion to 150 degrees and extension to minus 5 degrees, actuated by paired strings on servo-driven pulleys.

The sensing side includes a calibrated digital temperature sensor embedded in the spine behind a thin silicone layer, a custom resistive aeration sensor with gold-plated contact pads that detects air bubbles in the upper airway or lung cavity, and a turbine flow sensor in the umbilical circuit. Custom printed circuit boards route signals through the skeleton, and power, ground and data cables run through the umbilical cord to an external motherboard carrying an ESP32-S3 microprocessor, isolated DC-DC converters, an emergency stop, status LEDs and programmable buttons. The motherboard links to a mathematical model of fetal circulation, the digital twin, through a client-broker middleware called OOCSI, with all data stored in a research infrastructure called Data Foundry. In a demonstration scenario, occluding blood flow triggered modulation of cerebral saturation in the digital twin, which in turn changed the cyanotic expression on the physical manikin.

Realism was evaluated in a mixed-methods, cross-sectional study involving 53 healthcare professionals recruited from the gynecology and neonatology departments of Máxima MC in Veldhoven, with domain-specific subgroups of seven clinicians for cyanosis, eight for breathing, nine for limb movement, and seven functional plus 22 visual assessors for the umbilical cord. Questionnaires rated features on a 7-point Likert scale from completely unrealistic to completely realistic, supplemented by open-ended feedback and informed by an established fidelity framework. Ratings showed generally high perceived realism across domains, with medians of 5 to 6, and Friedman tests found no statistically significant differences among features within any domain. Cyanosis received the lowest overall ratings, with the dynamic appearance and disappearance of the blue discoloration judged less realistic than its color tone, although the prototype’s CIELAB values of L* 26, a* 16 and b* 17 approximated the clinical reference of L* 32.8, a* 14.8 and b* 15.2. The umbilical cord was judged more realistic than commercially available phantoms, particularly its pulsation and arterial spiraling, though assessors said its dimensions should be scaled down for a 24-week fetus.

The authors are candid about the limits of the work. The evaluation was exploratory and descriptive, with no a priori power analysis, and the small, non-normal samples called for non-parametric statistics. The blood circuit used a piston pump with a stroke volume of 1 to 3 milliliters at 148 beats per minute, deliberately higher than the 0.44 to 1.17 milliliters of an actual 24-week fetus, because the setup was adapted from an adult mock circulation, and the researchers note that a physiologically realistic pulsatile flow setup remains to be developed. The current prototype cannot simulate cannulation of the umbilical vessels to the artificial placenta, a critical procedure, and sensor density, feedback latency and skin tone fidelity all constrain realism. The team recommends future validation with larger, standardized instruments such as the Simulation Fidelity Rating scale, a comprehensive simulation exercising all functions together, and reproduction of preclinical animal studies on the manikin for additional verification.

Even so, the researchers see three roles for the hybrid manikin in artificial womb development. First, it can serve as a substitute patient for validating early-phase systems, measuring whether heart rate, blood flow and oxygen saturation remain steady during simulated procedures and exposing ergonomic problems and human errors before clinical trials begin. Second, once the design matures, it can train medical personnel in the womb-to-system transfer and monitoring in a controlled environment, quantifying proficiency as a function of task completion time while vital organ parameters are tracked. Third, it may eventually give parents and patient advocates a tangible, concrete experience of the treatment during counseling. The authors stress that clinical simulation alone provides only partial validation for novel treatments and that further validation in clinical settings is necessary before any clinical adoption, but they position the physical-digital twin approach as a way to reduce reliance on animal studies in early-phase research, defer human trials to a more predictable stage, and expand stakeholder involvement in high-risk, human-centered medical technology development.

Subject of Research: Development and realism evaluation of a hybrid fetal manikin connected to a digital twin for extra-uterine life support research

Article Title: Development of a hybrid manikin for extra-uterine life support technology innovation

Article References: van Haren, J. S., van Willigen, B. G., van Riet, C. M., Loor, W. J., van den Broek, J. R., Sterk, J., Amendola, C., Verschueren, K., Thielen, M. W. H., van der Hout-van der Jagt, M. B., & Delbressine, F. L. M. (2026). Development of a hybrid manikin for extra-uterine life support technology innovation. BMC Biomedical Engineering, 8(1), Article 17. https://doi.org/10.1186/s42490-026-00120-z

Image Credits: AI Generated

DOI: 10.1186/s42490-026-00120-z

Keywords: artificial womb, extra-uterine life support, fetal manikin, digital twin, medical simulation, extremely premature infants, artificial placenta, near-infrared spectroscopy, umbilical cord phantom, cyanosis simulation, biomedical engineering, simulation-based training