Electronic Health Records (EHRs) contain a wealth of patient medical information that can: save valuable time when an emergency arises; eliminate unnecesary treatment and tests; prevent potentially life-threatening mistakes; and, can improve the overall quality of care a patient receives when seeking medical assistance. Children's Hospital Los Angeles (CHLA) wanted to know if the records could be mined to yield early warning signs of patients that may require extra care or an indication of the severity of a patient's illness. In this lab we have access to the work and results of CHLA's applied use of deep neural networks on EHRs belonging to roughly 5,000 pediatric ICU patients.
We will use deep learning techniques to provide medical professionals an analytic framework to predict patient mortality at any time of interest. Such a solution provides essential feedback to clinicians when trying to assess the impact of treatment decisions or raise early warning signs to flag at risk patients in a busy hospital care setting.
In this lab we will use the python library pandas to manage the dataset provided in HDF5 format and deep learning library Keras to build recurrent neural networks (RNN). In particular, this lab will construct a special kind of deep recurrent neural network that is called a long-short term memory network (LSTM). Finally, we will compare the performance of this LSTM approach to standard mortality indices such as PIM2 and PRISM3 as well as contrast alternative solutions using more traditional machine learning methods like logistic regression.