I-Corps: Translation potential of prediction tool to determine physiological pregnancy due date

The broader impact of this I-Corps project is the development of a software tool for personalized pregnancy care. Currently, uncertainty in the ability to predict when labor will begin has significant implications for maternal morbidity and financial consequences for patients and health systems. Consequences range from reduced fetal/newborn survival for preterm to complex prolonged pregnancies, lack of timely access to appropriate levels of care (e.g., rural residents), extended hospitalization for false labor or labor induction, added strain on patients without social support needing to plan work leave or childcare, and higher morbidity and mortality for patients for whom labor is contraindicated.

Removing the guesswork around the onset of labor may improve the pregnancy experience and allow women to self-monitor their pregnancies, communicate with their care providers, and access the right level of care at the right time. This technology could provide a tool to monitor physiological change during pregnancy, especially between prenatal care visits, and guide decision-making helping patients as well as care providers across gestation stages.

The overall goal is to improve pregnancy outcomes by improving decision-making and addressing uncertainty.

This I-Corps project utilizes experiential learning coupled with a first-hand investigation of the industry ecosystem to assess the translation potential of an artificial intelligence (AI) model to monitor pregnancy and predict when labor will begin. The technology is designed to collect data from a non-invasive wearable sensor to measure skin temperature.

An autoencoder was trained using a long short-term memory (AE-LSTM) model on continuously measured temperature data from the skin using a smart ring. Data was used to predict the number of days until the onset of labor. Results show that across 37-42 weeks of gestation, the model predicted the onset of labor 7 days before the participants’ report of symptoms of labor.

The research confirmed that the predicted gestation and onset of labor aligned with actual gestational age at labor onset (linear fit of R2 of 0.93). In addition, predictions had an average error of <2 days from the labor onset. The goal is to develop a tool for predicting labor onset that could be employed by pregnant individuals or used by care providers in helping guide care.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.