Predicting ICU death with summarized patient data

Clinical Decision Support Tools (CDST), based on patients who are similar to the current patient might help a physician to deal with complex tasks. Many current researchers have focused on data mining and pattern identification with Electronic Health Records (EHR). CDST have been more popular as it captures significant trends and patterns in the data set without disclosing information of individual patients. In my project, I have focused on predicting ICU patient's mortality using clustered and un clustered data. I used MIMIC III online available database which includes information such as demographics, vital sign measurements made at the bedside (1 data point per hour), laboratory test results, procedures, medications, caregiver notes, imaging reports, and mortality (both in and out of hospital). The primary focus was on how the inputs (e.g. saline) and outputs (e.g. blood, urine, chart values) variables along with the LOS (length of stay) and weight of the patient can predict the death state for a particular ICU patient.