We aimed to predict infant mortality and survival time from electronic health records. We collected 11 years of delivery data from one women’s hospital and birth/death certificates from a county office. We developed data-driven predictive models that provided mortality risk and survival time. Their AUCs ranged 0.88-0.94. To our knowledge, we are the pioneers merging predictive modeling and survival analysis. Our models can potentially identify moms or infants with high-risk infant mortality in practice.
Learning Objective: Critically analyze infant mortality prediction and survival time estimate
Rich Tsui (Presenter)
Children's Hospital of Philadelphia
Luge Yang, Children's Hospital of Philadelphia
Lingyun Shi, Children's Hospital of Philadelphia
Lezhou Wu, Children's Hospital of Philadelphia