Location:Home > Engineering science > Computer Science > Applied Computer Technology > Bayesian network model of medical quality mining research

Bayesian network model of medical quality mining research

Downloads: []
Tutor: ZhaoZheng
School: Tianjin University
Course: Applied Computer Technology
Keywords: Quality of care,Bayesian networks,Structure Learning,MedicalK2,Reliability Analy
CLC: TP399-C8
Type: Master's thesis
Year:  2009
Facebook Google+ Email Gmail Evernote LinkedIn Twitter Addthis

not access Image Error Other errors

Medical record is a centralized medical record summary content, which includes the basic situation of inpatient discharge diagnosis, rescue, hospital costs and other information, is to carry out medical quality assessment of vital information. At present, most hospitals use statistical methods to analyze them, but only derive the specific values ??of medical quality indicators, the data can not be implied in the quality of care issues more expression. With the dramatic increase in the medical record data, the performance of traditional methods increasingly difficult to meet the needs of hospital management. Affect the quality of care indicators in order to find the value higher or lower than the expected value of all possible factors, the paper medical record data for the study, the medical quality indicators as a starting point, through the use of Bayesian networks for medical record data, a potential causal relationship excavation and analysis, in order to achieve the purpose of improving the quality of hospital care. The main work is as follows: (1) propose a Bayesian network structure learning algorithm - MedicalK2. The greedy search algorithm and Bayesian score as the core framework, combined with a randomly generated sequence supplemented by other constraints nodes together to complete the Bayesian network structure learning task. Through a large number of experimental results, the validation of the algorithm on the data in the validity of the medical record. (2) based on the use of medical quality indicators to build Bayesian network modeling strategy. For the medical record contains data with information about health care quality indicators proposed in this paper based on the quality of care indicators build a Bayesian Network model research methods. As a graph structure, Bayesian networks can well reflect the different characteristics of a causal link between, and in the form of probability describe the strength of this link, to facilitate evaluation of medical quality is good or bad. (3) discuss the Bayesian network model reliability. In order to assess the correlation in this study to build Bayesian network model adequately reflects the data in the medical record potential causal relationship, we propose a Bayesian network reliability evaluation method: random with replacement sampling strategies to obtain a certain amount of sample space capacity similar new data set, using MedicalK2 modeling, and the number of occurrences of each side statistics to determine if the network is reliable. Based on the above, this paper proposes a set of data based on the medical record, and tap the potential causal medical quality integrated solutions, including the initial medical record data processing, selection and implementation of research methods, experimental results of the analysis and summary, and through Tianjin, a hospital medical record four consecutive years of data mining results demonstrate the feasibility and effectiveness of the solution.
Related Dissertations
Last updated
Sponsored Links
Home |About Us| Contact Us| Feedback| Privacy | copyright | Back to top