| In the emergency medical process, how to reduce the avoidable adverse events, how to guarantee the security of patients and how to improve efficiency and quality of medical treatment from the perspective of management science are always the key subjects in theory field and practical affairs field. There is some deficiency in the existingresearches on reasoning mechanism of CBR/RBR fusion and the influence of uncertainty on the reasoning robustness of medical decision-making in heart disease. In this dissertation, the robustness principle was introduced to study the robustness in the reasoning of medical decision-making. The innovative work is as follow:1. The robustness principle was introduced to the study on the quality and efficiency of medical decision-making in heart disease (MDMHD). Aiming to the low reasoning efficiency and accuracy of MDMHD, the robustness principle was firstly adopted into MDMHD. At the same time, the conception of robustness of MDMHD was proposed in the dissertation. The bottlenecks of MDMHD, including anti-interference, fault tolerance, and redundancy, were studied. The rule on acquisition and transfer of knowledge was systematically discussed, which reveals the key role the knowledge reasoning plays. A group of robustness criterion was built up by the method of robust thresholds, and the mechanism of fusion coordination was built through information entropy, information gain and mutual information. A set of fusion models and robustness threshold method were proposed, including R2CMIFS and RTCRF, which enriched the theory of robustness in this field.2. Robustness threshold was adopted to solve the fusion reasoning of CBR/RBR. The concept of unitary space in fusion reasoning and the vector of robustness thresholds were first proposed in the research on the fusion mechanism of CBR/RBR during MDMHD. For robust optimization of fusion system, the singular value decomposition (SVD) in matrix theory was implemented to construct the threshold vector of CBR/RBR fusion algorithm. The vector was employed to be the constraint of fusion reasoning, by which the boundaries of the fusion reasoning problem were determined and the uniqueness of the resolution of the fusion reasoning problem was proved. Thus, thismethod handled off the vulnerability of the systems in decision-making fusion, which was superior to the fusion mechanism proposed by LUENGO, et al. . Based on the strategies of resource separation and the mechanism of conflict elimination, CBR/RBR fusion model was built in the mode of collaborative fusion. Therefore, the deficiency of standard robust objective function introduced by BEN-TAL, et al. was improved, by effectively resolving the uncertainty of the heterogeneous information in the proposed model.3. The reasoning method was proposed to deal with implicit and unstructured knowledge in uncertainty decision. Aim to the study on decision-making of heart disease first aid under uncertainty, a compound and superposition model was proposed which consist of CBR/RBR stable model and Random item in reasoning process. Thefusion reasoning mechanism of CBR/RBR based on the Bayesian (BN- CBR/RBR) was established. From the view of accuracy and convergence, BN-CBR/RBR model showsits advantage of robustness, in comparing with CBR/RBR stable model proposed in foreign literatures. As a result, the medical decision-making theory under uncertainty was enriched. The core function of fusion problems on implicit and unstructured knowledge in decision-making under uncertainty was discussed. Dual mechanism and collaborative optimization theory were applied to establish BN-CBR/RBR optimization method based on multi-attribute decision-making. The researches on the application of human-computer in medical decision-making were expanded.4. The theory and model of robustness was applied to the decision-making in heart disease. The proposed model and method of robustness solved the uncertainty in the decision-making system efficiently, and great innovations and breakthroughs were madein enhance the robustness of medical decision-making system. Through the embedded algorithm, the model and method can be cured in prototype system to provide the assistant decision-making for MDMHD. Consequently, the contradiction between the lack of medical expert sources and their low utilization can be coped with, which is significant for improving medical quality and efficiency in decision-making and plays a key role in constructing the digital medical system with international standard.This dissertation is supported by Tianjin Research Program of Application Foundation and Advanced Technology (Grant No.10JCYBJC07300), Tianjin Municipal Science and Technology Commission (Grant No. 09ECKFGX00600) and FOXCONN Group (Grant No. 120024001156)... |
PDF Full Text Request