Data Sources and Aggregation


A challenge in neurocritical care is the aggregation of data from multiple diverse sources so that it can be used efficiently in patient management as well as in algorithms that extract additional value. The data of interest is often in silos and may not be in the most usable form. Additionally, the interoperability of data from source to user has been challenging. Four sources of data are being addressed:

  1. Medical device data
  2. Electronic medical record data
  3. Imaging
  4. Nursing interventions

1. Data Aggregation - Medical Devices

Our focus is on the capture of high-resolution physiological data in neurocritical care as this is one of the most significant gaps in creating a connected neuro ICU.  There is no widely used medical device communications standard which has led to these gaps.  Some devices send waveform and trend data over the hospital network and several commercial products can collect this data (Capsule Technologies, Excel Medical, Sickbay). Some do it for the purpose of sending the data to the medical record but it may end up in a resolution that might not be useful for precision management.   Others have developed systems that collect the high resolution data directly from devices (Moberg Research, ICM+).  


Our approach is not to attempt to develop a medical communication standard due to the time and effort involved.  Other groups are working on such a standard.  Rather our approach is to interface to devices or to devices that already aggregate signals and convert the data into a common format.


Our Group has developed recommendations for medical device manufacturers that guide them in the development of a robust communications protocol.  We have presented our work at meetings and published it in journals. Current work is partnering with clinical trials and developing software that can robustly  See Work Products.

2. Data from the Electronic Medical Record

EMR data is important to provide context to the physiological data and vice versa.  Unfortunately, EMR data has been elusive to obtain, however, that is changing with the advent of FHIR (Fast Healthcare Interoperability Resources) and other technologies. 


Our approach is to partner with other organizations addressing automated communications with the medical record systems.  We are using manually retrieved EMR data to demonstrate the contextual value of EMR data (such as medications) to the physiological data.  


We have developed upload methods to get the data to our cloud storage as well as methods to harmonize the data from two centers both using Epic. We have developed displays to show the time synchronization of medications with the physiological data.

3. Imaging

Image data is important in the management of patients in neurocritical care and it can be obtained via a PACS viewer using the DICOM standard.  Our focus will be to investigate image descriptors that can be used in the analysis of the data.  We have not worked on this project yet.

4. Nursing Interventions

Nursing interventions are generally recorded in the EMR but this record may not be complete  In addition, the synchronization of time from when the event occurred to when it is noted in the record could be quite variable. We will investigate methods to address these issues.