GRU-D Characterizes Age-Specific Temporal Missingness in MIMIC-IV

Temporal missingness, defined as unobserved patterns in time series, and its predictive potentials represent an emerging area in clinical machine learning. We trained a gated recurrent unit with decay mechanisms, called GRU-D, for a binary classification between elderly - and young patients. We extracted the first 24h of patients' time series for 5 vital signs from MIMIC-IV as model inputs. GRU-D was evaluated with means of 0.778 AUROC and 0.797 AUPRC on bootstrapped data. Interpreting model parameters, we found differences in temporal missingness of blood pressure and respiratory rate learned by parameterized hidden gated units. We successfully showed how GRU-D can be used to reveal patterns in temporal missingness potentially building the basis of advanced imputation techniques.