APPEAR - Analyse von Patientenbehandlungs-Prozessen zur Erkennung von Abweichungen und Regelmäßigkeiten

One challenging question that medical staff is facing day by day is, whether their patients are treated according to available clinical pathways or whether patients could have been treated better. Answering this question is time-consuming and an IT-solution that supports this process would be desirable. For the development of such a solution, two engraving problems have to be solved:

  • In order to compare the individual treatment process of a given patient with an electronically available and processable representation of a clinical pathway, an all-embracing overview about the treatment of the patient must be available. This overview cannot be gained completely by an IT-solution, because the necessary pieces of information are mostly scattered over a rich set of paper-based as well as of electronic sources. Thus, the information that is only available on paper is not on-hand for the system and is therefore "lost" for the comparison. But even the electronically available information can often not be used without further analysis: most hospitals run a variety of IT-systems, e.g. IT-systems for administrative tasks, IT-systems for storing findings or IT-systems for storing surgical reports. Each of these systems contains information that is necessary to receive an overview about the treatment of a patient, but mostly these systems are only poorly integrated and often include contradicting information.
  • The second challenge is the acquisition of electronically available and processable representations of clinical pathways. Modelling them by hand is very time consuming and hard to do for health professionals and computer scientists: health professionals often have difficulties in formalizing typical processes and computer scientists lack the necessary medical knowledge.

These problems have been tackled during the development of the Appear-System (Analyse von Patientenbehandlungs-Prozessen zur Erkennung von Abweichungen und Regelmäßigkeiten). The Appear-System bases on the Collaborative Agent-based Knowledge Engine, a general framework that can be used as a basis to build individual support systems and which is currently applied in several application domains. The Appear-System is a prototypical implementation, tailored for the IT-infrastructure of the University Hospital in Greifswald. The considered group of patients are stroke patients, especially patients with an ischaemic stroke, an intracerebral bleeding and an subarachnoidal bleeding.

The development of the first prototype of the Appear-System is finished. The user-interaction is realised as an HTML interface, so that each user within the network of the University Hospital in Greifswald can easily use the system. It provides the following functionalities:

  • Appear is able to query different IT-systems that provide information about an individual patient and generates a workflow instance for this patient that collects all electronically available information. The user can navigate through the workflow instance and receives all electronically available information within one central view.
  • Appear is able to compare a workflow instance of an individual patient with a corresponding clinical pathway and presents all inconsistencies and abnormalities to the user.
  • Appear is able to generalize several workflow instances to form a new clinical pathway. Therefore, the workflow instances are compared and similar sub-workflows are identified. The result can than be used as a basis to refine it by hand.

Research Team

The Appear-System is developed in collaboration with the University Hospital in Greifswald, especially with the Clinics and Polyclinics for Neurology.

Two members of the Wi2-group are working within Appear.

 

Six members of the University Hospital in Greifswald have intensively been involved in the knowledge acquisition step.