A Cox model with functional principal component scores to predict insulin initiation in women with gestational diabetes

Yueyun Zhu, University of Galway

Co-authors: Fidelma Dunne, University of Galway; Christine Newman, University of Galway; Robert McEvoy, University of Galway; Andrew Smyth, University of Galway; Mary Fitzgerald, University of Galway; Andrew J. Simpkin, University of Galway

Abstract: Gestational diabetes mellitus (GDM) is a global health problem. To manage glucose levels, insulin is recommended when pharmacotherapy is required. We aim to predict insulin initiation in women with GDM. A landmark approach with Cox models is applied. Functional principal component analysis (FPCA) is employed to summarise the glucose trajectories using FPC scores. A Cox model with FPC scores demonstrates strong performance in predicting time-to-insulin initiation.