Wearable devices are advancing in the healthcare sector and with their most recent application it is focusing on the prediction of preterm birth risk. A study conducted at Stanford University illustrates how the utilization of machine learning on data gathered from wearables throughout pregnancy can provide insights into the enigma surrounding premature births.
The study was published in the latest edition of the Digital Medicine and it stands out due to its extensive and diverse participant group. Over 1,000 women, with more than half of them being Black, were monitored throughout their pregnancies by researchers from Washington University in St. Louis.
These expecting mothers wore motion- and light-sensing watches throughout their pregnancies. Their continuous data was analyzed alongside their medical records using deep learning techniques. Surprisingly, the machine learning model could estimate how far along a woman was in her pregnancy with reasonable accuracy based on her activity and sleep patterns.
Furthermore, the study revealed a detailed observation like certain women exhibited pregnancy progress beyond the algorithm’s predictions, as evidenced by disturbances in their activity and sleep patterns, and astonishingly, these individuals were identified as having a 44% increased likelihood of undergoing preterm births. It is important to note here that the research does not suggest a direct causal relationship between reduced activity or disrupted sleep and preterm births but rather presents a hypothesis to be investigated in future studies.
This innovative approach holds the potential to create early warning systems that can aid healthcare providers in identifying high-risk pregnancies. Additionally, it provides optimism for unraveling the complex factors contributing to racial disparities in preterm birth rates and this has been a longstanding puzzle for researchers. Jessica Walter, a reproductive endocrinologist, said that pregnancy serves as an ideal research window.
