Johns Hopkins researchers are tapping into the rich data of bedside monitors to create algorithms aimed at improving the outcomes of neurocritical care patients.

When patients are in the neurocritical care unit, the host of devices used to monitor them, such as mechanical ventilators and intracranial pressure monitors, together produce over 250 data points at any given time. Some devices record the data, but the information is usually only accessible for doctors and nurses to review for a limited number of hours while the patient is being monitored. Once a patient leaves the neurocritical care unit, most of the data is lost, with only fragments saved.

Jose Suarez — director of the Neurosciences Critical Care Division and professor of anesthesiology and critical care medicine, neurology and neurosurgery — and his colleagues are capturing deidentified data from hundreds of patients to develop an algorithm that recognizes patterns in patients who experience worsening in neurological function. The team will also develop a separate algorithm that predicts which patients will require intensive care as opposed to less observation in a nonintensive care unit.

“This is precision medicine for the neurocritical care unit,” says Suarez. “Automated, early detection of crises precursors can help doctors intervene to prevent secondary brain injuries.”

In July 2018, Suarez began using a data-gathering platform to continuously collect all of this real-time information in the neurocritical care unit. Between 7 and 10 terabytes of data will be stored annually on servers at Johns Hopkins’ Mount Washington data centers. Suarez and Peter Dziedzic, research associate and Center of mHealth and Innovations director, will then use these data to build the algorithms.

The team expects the algorithms will improve not only patient outcomes but also patient flow and lessen patient time spent in the hospital. Testing is planned for 2019; in the meantime, Suarez is working to identify additional funding sources.