Fig. 1
From: A deep learning model for real-time mortality prediction in critically ill children
Prediction trajectory using PROMPT. Serial trends of recorded vital signs during ICU stay and hourly calculated predicted mortality rate using PROMPT shown for patients from the validation cohort who survived (a) and died (b), respectively. Predicted mortality was averaged over multiple prediction models trained from development cohort using the dataset to predict mortality in the preceding k hours (where k = 6, 12, 24, 48, and 60). SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure; HR, heart rate; RR, respiratory rate; SpO2, peripheral capillary oxygen saturation; BT, body temperature; ICU, intensive care unit