Mining Vital Signs Time Series to Improve Clinical Outcomes for Traumatic Brain Injury Patients
Posted on October 13, 2011
We have been awarded a Collaborative Research Experience for Undergraduates (CREU) grant from the Computing Research Association Committee on the Status of Women in Computing Research (CRA-W) and the Coalition to Diversify Computing (CDC) for a project entitled: "Mining Vital Signs Time Series to Improve Clinical Outcomes for Traumatic Brain Injury Patients". This is a joint grant with Tim Oates (UMBC) and Peter Hu (R Adams Cowley Shock Trauma Center). The grant will fund two research students at Wheaton and two at UMBC plus student conference travel.
Baltimore's Shock Trauma Center sees about 8000 patients per year, with many suffering from traumatic brain injury (TBI). These patients run the risk of secondary injury due to swelling of the brain and resulting increased intra-cranial pressure (ICP), which is currently measured invasively by drilling a hole in the skull for a sensor. We are interested in developing non-invasive proxies for ICP based on high frequency vital signs data recorded non-invasively for all Shock Trauma patients. The goal is to apply machine learning algorithms to estimate current ICP using vital signs, and to predict the need for life saving interventions such as blood transfusions or surgery. If successful, our research can lead to better care and long-term outcomes for TBI patients.