Medical Diagnosis Based on Multi-threshold Fuzzy Petri Net with Loop

To settle the loop judgement issue of the medical diagnosis process, a multi-threshold fuzzy Petri net with loop (MTFPNL) is proposed to simulate the evidence-based medical diagnosis process of“ initial diagnosis-confirmation”. The utilized mechanisms of MTFPNL could be classified into the following aspects. Firstly, a novel multi-threshold transition is presented to make a judgement based on the doctors ′ examination data at the initial stage and to select the next step in the diagnosis process. Then, a loop structure is employed to conduct the multiple assisted medical diagnosis when the diagnosis cannot be confirmed by existing data and to strengthen the model ′ s selection of the diagnosis. Finally, an inference algorithm based on the presented MTFPNL is discussed to perform the evidence-based medical diagnosis process automatically. To further verify the feasibility of the proposed model, a set of cervical spine disease data is utilized to validate the model ′s ability. The experimental results reveal that the MTFPNL can effectively describe and execute the evidence-based medical diagnosis process of“initial-confirmed”. Meanwhile, this formalism also expands the application domain of FPN from acyclic FPN to looped FPN.