EMERGENCY NEUROPHYSIOLOGICAL ASSESSMENT BEDSIDE LOGIC ENGINE -
DESCRIPTION (provided by applicant): Emergency Neurophysiological Assessment Bedside Logic Engine We propose to develop an intelligent clinical informatics search tool that is integrated with automated brain monitoring with dense array EEG (dEEG; 128 or 256 channels). Many forms of neurological injury, such as hematoma or nonconvulsive seizure, are difficult to diagnose in the emergency room, and yet this is the point at which they are often most effectively treated (Jordan, 1999). Failure to recognize treatable brain injury leads to extensive suffering for patients and families and costs to society, such that an inexpensive automated brain monitor could be highly cost-effective in the emergency context. Continuous monitoring of brain function can now be accomplished easily and inexpensively in the emergency setting with dEEG (Luu, et al., 2001), providing key neurophysiological information for emergency neurological assessment (Jordan, 1999, Procaccio et al., 2001). There are, however, both technical and professional challenges to make dEEG monitoring practical for routine use. Technically, continuous monitoring is required for many neurological conditions, and this is not practical with visual inspection of EEG traces or crude automated measures such as integrated amplitude. Reliable automated pattern recognition is required, yet in the past this has often failed to separate real neurological disorder from artifacts such as movement, eye blinks, and cardiac signals. There are also professional challenges, one in acquiring the brain data and another in interpreting it. First, EEG technologists are often not available in the emergency department, so that EEG sensor application must be easy enough for nurses and medical assistants who have completed a short, focused training protocol. Second, emergency physicians are not trained in EEG interpretation so they require assistance. Optimally, this would both access to consultation by an expert neurologist and an intelligent search engine that reads the dEEG pattern results and provides an evidence-based interpretation in relation to diagnostic questions to be evaluated. We propose a two-year research and development program to meet both these technical and professional challenges. We will begin with a intelligent search tool, the Clinical Decision Analysis (CDA) system from Lifecom, Inc. that has been proven effective for providing online guidance to a physician or physician's assistant in the emergency department. The CDA helps with evaluating symptoms in the context of medical evidence, hypothesizing disease states, ordering lab tests, and making diagnostic and treatment decisions. An extensive database of clinical evidence and guidelines is made available at the bedside, with contextual search tools that track the stages of evidence-gathering and clinical decisions required for emergency evaluation. We will develop a specialized version of the CDA for evaluation of emergent neurological disorders, and we will provide evidence on neural status from online monitoring of dEEG. Rapid (one minute) application of the dEEG sensor net by the first responder allows monitoring to begin when the patient is first contacted. Pattern recognition with the dEEG data is provided by high performance compute clusters: (1) to separate brain signals from noise (e.g., movement, cardiac, equipment artifacts) and (2) to recognize pathological brain states (e.g,. seizure, burst-suppression coma, drug toxicity, vasospasm, focal slowing due to hematoma). The fusion of automated neurophysiological monitoring with the intelligent diagnostic search tool will create the Emergency Neurophysiological Assessment Bedside Logic Engine (ENABLE). ENABLE can be seen as a new paradigm in which informatics allows a two-way exchange between clinical data gathering and clinical data interpretation. In supporting clinical data gathering, the knowledge assembled by an intelligent search tool (such as the clinical presentation of a patient with mild head trauma, together with expected complications such as hematoma) is used to provide a predictive analysis of the brain monitoring data, highlighting the patterns that may be most diagnostic in that context (such as the progression of regional EEG slowing that may signal a developing hematoma). Thus, the difficult challenge of pattern recognition is made easier by placing it in the continually developing diagnostic context framed by the informatics logic engine. The improved pattern recognition of neuropathology then paves the way to improved clinical decision making. Through the fusion of automated neurophysiological pattern recognition, ENABLE not only summarizes the patient's brain state for the physician, but it presents a set of rule-out and rule-in protocols that allow active hypothesis-testing, and additional data gathering, to inform the diagnostic decision. Instead of an intelligent search tool that begins only with the physician's queries and observations, ENABLE can act in the background to evaluate the patient's ongoing changes in neurophysiological state and to assemble an intelligent set of diagnostic options that are consistent with those changes. These are tested against the additional diagnostic and clinical data available on that patient, such that ENABLE can then set alarms, launch literature searches, suggest additional tests, and otherwise assist the diagnostic and treatment processes. Building ENABLE requires extending the CDA Knowledge Repository (KR) to recognize neurophysiological pathology in the dEEG signals and to link this pathology to neurological disease states and diagnostic syndromes. Our neurology consultant, Dr. Mark Holmes, is a clinical neurophysiologist who will work with Dr. Datena and other emergency physicians to extend the CDA KR to emergent neurological conditions. Given the capture and formulation of this clinical knowledge within the KR, we will turn to a key goal for the widespread adoption of ENABLE: a program of simulations that train medical personnel in emergency neurological evaluation, including integrated dEEG pattern recognition when this is available. Through this integration, ENABLE will offer the opportunity for a new form of intelligent clinical search tool that is linked directly to an advanced technology for automated physiological monitoring.
PUBLIC HEALTH RELEVANCE: This project would create an advanced computing technology for continuous brain monitoring in emergency and intensive care settings. A brain wave sensor net allows rapid application of a dense array of electroencephalographic (EEG) sensors with elementary training. Advanced computation methods allow automated detection of seizures and other forms of injury that put the brain at risk. The complete system will be integrated with clinical decision assistance software that will provide emergency physicians with guidelines for incorporating the brain status information into immediate medical decisions, and for seeking expert neurological review when necessary. Network access technology will facilitate remote review by expert neurologists, and a training protocol will provide first-responder usability for emergency technicians and physicians without training in clinical neurophysiology.
