Fetal movement is a crucial indicator of the proper development of the fetal nervous and musculoskeletal systems. However, approximately 15% of pregnant women experience reduced fetal movement, which is often associated with fetal death or stillbirth. A healthy fetus typically cycles between activity and quietness every 30 to 40 minutes, while a compromised fetus may reduce movement to conserve oxygen—a change that may occur before stillbirth. Therefore, accurately identifying fetal movement patterns is critical. Yet, current monitoring methods remain relatively limited: the most common approach relies on maternal self-perception, but studies show that about 40% of fetal movements go undetected; although ultrasound can visually observe fetal movement, it requires professional operation and is confined to hospital settings; fetal heart rate monitoring devices are bulky, necessitating straps for fixation and require specialized expertise for operation and interpretation, so they are generally reserved for high-risk pregnancies. It can be said that truly convenient, objective, and practical fetal movement monitoring technology remains a clinical gap to this day.

Recent studies have proposed several new methods for fetal movement monitoring, including detecting pressure changes caused by fetal movements using piezoelectric sensors, capturing micro-vibrations generated by fetal movements with micro-mechanical acoustic sensors, and monitoring maternal abdominal displacement through accelerometers. However, these methods generally suffer from severe signal interference issues, such as environmental noise and vibrations masking the fetal signals themselves; accelerometers struggle to distinguish between maternal movements and fetal movements, and require strict positioning for accurate readings, otherwise accuracy significantly declines. Additionally, most of these wearable devices necessitate bulky straps for use, resulting in poor wearability and limiting their daily practicality.

This study presents an AI-powered, adhesive patch-type wearable pressure-strain composite sensing system capable of continuously and accurately capturing minute maternal abdominal skin displacements induced by fetal movements. The octagonal gold nanowire film designed in this research functions as a omnidirectional strain sensor, covering approximately 77 and 217 cm² in two-dimensional and three-dimensional biomimetic abdominal models, respectively. In contrast, the pressure sensor, while offering higher local detection precision, has a smaller effective range of only about 13 and 38 cm². Leveraging the characteristics of both sensor types, this study further developed a highly integrated adhesive patch with a thickness of ~3 mm, an area of 10–14 cm², and a weight of ~3 g. In tests involving 59 pregnant women, data collection was achieved using just two patches, and a machine learning model achieved over 90% fetal movement recognition accuracy (verified by ultrasound). This intelligent, lightweight, and comfortable wearable patch-based sensing system demonstrates application potential for continuous fetal movement monitoring outside of clinical settings.

Source: Sensor Expert Network