The Design Of Clinical Pathway Based On Process Mining

Clinical pathway (CP) is an advanced medical service management model, in China, some large hospitals are trying to apply it. Clinical pathway management includes prepare, planning, executing, improving, supervising and evaluating, the planning of Clinical pathway is the basis of the whole CP management. In this thesis some research is done in the scientific design of Clinical pathway (CP), and a method based on process mining was proposed for the design of CP.Firstly, recent algorithms of process mining are introduced, and the heurist process mining algorithm dealing with the noise data in the log proposed by Aalst is revised considering the specialty of diagnosis activities in hospital. Judge the relationships between diagnosis activities by comparing the metrics measuring the relationship and the threshold that settled, then, build the diagnosis flow based on the relationships, and the flow net is described in workflow model.Secondly, the mined diagnosis flow model is evaluated. The metric"fitness"is to measure the fitness between the log and the mined flow model. Check one activity at one time in the every diagnosis flow orderly, see if the activity can be executed according to the flow model , if not , the"fitness"will be given an punish value.Thirdly, deal with the uncertainty of the threshold value in the heurist process mining algorithm, a method based on design of experiment (DOE) is introduced. Take the threshold as variable and the fitness of the flow model as the response, the optimal combination of the threshold value is calculated through DOE analysis.Lastly, the diagnosis flow mining system is developed based on the workflow mining method. The method is then used in developing CP of cesarean section and its effectiveness has been validated. The results demonstrate that the method---use DOE to calculate the optimal threshold value, then use this threshold value to mine the diagnosis flow through heuristic rules can mine an appropriate flow model. And the diagnosis flow model can provide effective decision support for the development of clinical pathway.