Persistent entropy links irregularities in daily and weekly rest and activity cycles during gestation week 22 and 32 to maternal and neonate health outcomes: A prospective cohort study

Though rest (i.e., sleep and periods of inactivity) and activity levels are known to be altered during pregnancy, the effect of day-to-day consistency in rest-activity cycles on maternal-neonate health is not well characterized. Actigraphy objectively measures rest-activity; however, the large datasets it generates often reveal irregular patterns that are challenging to interpret. We introduce a novel approach that applies persistent homology and entropy, quantitative measures in the field of topological data analysis, to quantify disorder in actigraphic data and examine their association with maternal and neonate health. Actigraphic data were collected from a prospective observational cohort study of pregnant women, with observations at gestational weeks 22 (G22; n = 41) and 32 (G32; n = 44). Persistent entropy was computed for daily and weekly actigraphic data. Participants who experienced maternal or neonate complications, including gestational hypertension, preeclampsia, etc., and adverse birth outcomes had higher average daily entropy at G22 and G32, indicating greater temporal disorder in rest-activity. Furthermore, complicated pregnancies had higher odds of exhibiting high entropy compared to uncomplicated pregnancies OR = 10.38, 95% CI 2.03, 74.68; Fisher’s exact test, p = 0.001), especially in participants with high BMI (OR = 24.00, 95% CI 1.76, 1549.19; Fisher’s exact test, p = 0.005. Receiver operating characteristic (ROC) analysis showed that weekly entropy at G22 had acceptable predictive power (AUC = 0.70; threshold = 8.24; sensitivity = 0.86; specificity = 0.70), whereas entropy at G32 was a weaker predictor. While the analysis yielded a significant association between high entropy at G22 and complicated pregnancies, the wide 95% CI 1.76, 1549.19 suggests that the magnitude of this effect should be interpreted with caution due to the limited sample size. These findings suggest that persistent entropy may be a useful tool for linking rest-activity irregularities to maternal and neonatal health outcomes. Larger studies are needed to evaluate its potential as a predictive model.