A Data Analytics Framework for Improving the Efficiency of Stroke Imaging Investigations

Emergency departments are under tremendous pressure to provide high-quality care in the shortest amount of time possible. While not all cases seen in the ED are urgent in nature, some require immediate attention. These true emergencies are usually complex in nature and depend on people, processes and technologies to work seamlessly, in perfect orchestration, in order to achieve the desired outcomes for the patient. One such condition is the stroke, a condition which left untreated (or treated incorrectly) can lead to devastating debilities and even death. To treat stroke successfully, a correct imaging diagnostic needs to be placed and treatment initiated within 3 hours of symptoms onset. Few other conditions in medicine have more stringent time requirements and fewer depend more heavily on timely imaging results for treatment decisions than stroke. Current guidelines mandate that a radiology report for suspicion of stroke be available within 25 to 30 minutes of patient arrival at the institution. As a stroke center of excellence, MU aims to consistently meet these guidelines. However, the complexities of the condition itself and of the system of care suggest that variation will occur. The purpose of our project is to assess the suitability of informatics tools like process mining and statistical process control to study the efficiency of imaging in the stroke care, to assess variability and its sources and to facilitate interventions to increase efficiency and reduce variability.