THURSDAY, Sept. 28, 2023 (HealthDay News) — Deviations from a normal “clock” of physical activity and sleep during pregnancy are associated with pregnancy outcomes, including preterm birth (PTB), according to a study published online Sept. 28 in npj Digital Medicine.
Neal G. Ravindra, Ph.D., from the Stanford School of Medicine in California, and colleagues used physical activity data from a wearable device, including more than 181,944 hours of data from 1,083 participants, to examine associations between physical activity and sleep patterns during pregnancy and PTB. A “clock” of healthy dynamics during pregnancy using gestational age as a surrogate for progression of pregnancy was developed using a new deep learning time-series classification architecture. Novel interpretability algorithms that integrate unsupervised clustering, model error analysis, feature attribution, and automated actigraphy analysis were developed, allowing for model interpretation with respect to sleep, activity, and clinical variables.
The researchers found that for modeling the progression of pregnancy, the model performed better than seven other machine learning and artificial intelligence methods. There was a strong association seen for deviations from a normal “clock” of physical activity and sleep changes during pregnancy with pregnancy outcomes. There were 0.52 fewer PTBs than expected and 1.44 more PTBs than expected when the model underestimated and overestimated gestational age, respectively. There was a negative correlation observed for model error with interdaily stability, indicating that when the daily rhythms were less precise, the model assigned a more advanced gestational age.
“This is exciting preliminary data,” coauthor Nima Aghaeepour, Ph.D., from Stanford University, said in a statement. “The results suggest that scientists should run studies to test whether tracking and modifying pregnant women’s sleep or physical activity could their lower prematurity risk.”
Copyright © 2023 HealthDay. All rights reserved.
Create Post
Twitter/X Preview
Logout