Computer Visualisation of Interrelationships Between Multiple Variables Across Human Pregnancy

Preterm birth refers to babies born before 37 weeks. Though there now some promising developments, in the last 30 years little progress has been achieved in preventing preterm birth, or in identifying those women likely to give birth prematurely. Our project focuses on preterm birth prediction. Specifically, this paper introduces a PBP (preterm birth predictor) program to reveal relationships between serial blood measurements of hormones and preterm birth. PBP uses 8 types of blood hormone data, namely, corticotrophin releasing hormone (CRH), human chorionic gonadotropin (hCG), alpha-fetoprotein (AFP). Estriol (E3), progestrone (Prog), Estradiol (E2), alkaline phosphatase (ALP) and Cortisol. Preliminary analysis shows that the actual CRH preterm median trajectory is higher than median term and median post-term trajectory. Similar results are also found for hCG, AFP and cortisol. These results suggest that multiple serial hormone measurements may offer a way of identifying a significant proportion of women likely to give birth prematurely.