Using Clinical Notes with Time Series Data for ICU Management

摘要: Monitoring patients in ICU is a challenging and high-cost task. Hence, predicting the condition of during their stay can help provide better acute care plan hospital’s resources. There has been continuous progress machine learning research for management, most this work focused on using time series signals recorded by instruments. In our work, we show that adding clinical notes as another modality improves performance model three benchmark tasks: in-hospital mortality prediction, modeling decompensation, length forecasting play an important role management. While time-series data measured at regular intervals, doctor are charted irregular times, making it to them together. We propose method jointly, achieving considerable improvement across tasks over baseline model.