声明
Chapter 1 Introduction
1.1 Background and Motivation
1.1.1 Epidemic Status of Diabetes Prevalence
1.1.2 Data Mining Strategies in Decision Making
1.2 Research Objective, Significances, and Innovation
1.2.1 Research Objective
1.2.2 Research Significances
1.2.3 Research Innovation
1.3 Current Issues and Challenges
1.3.1 Data Handling and Integration Issues
1.3.2 Data Cleaning Issues
1.3.3 Challenges in Preprocessing
1.3.4 Challenges in Model Selection
1.3.5 More Challenges in Medical Data Mining
1.3.6 Primary Health Education Challenge
1.4 Main Content and Summary of Research
Chapter 2 Basic Theories and Methodological Strategies
2.1 Basic Theories of Data Mining in Healthcare
2.2 Methodological Strategies for Diabetes Prevalence
2.3 Summary
Chapter 3 Regression Modeling for Group Covariates Correlation among Diabetes Features by Fractional Polynomial Approach
3.1 Introduction
3.1.1 Problem Statement
3.1.2 Shortcomings of Existing Methods
3.1.3 Proposed Method and Significances
3.2 Methodology and Preprocessing
3.2.1 Data, Framework, and Attributes
3.2.2 Logistic Regression Modeling
3.2.3 Fractional Polynomial Approach
3.2.4 Model Suitability
3.4 Experimental Results
3.4.1 Assessment of “Age and Occupation”
3.4.2 Assessment of “Occupation and Age”
3.4.3 R Correlation Plot
3.5 Discussion
3.5.1 Comparison with Cox Regression Method
3.5.2 Final Comments
3.6 Summary
Chapter 4 An Euclidean Grouping and Clustering Approach for Diabetes Features
4.1 Introduction
4.1.1 Problem Statement
4.1.2 Shortcomings of Existing Methods
4.1.3 Proposed Method and Significances
4.2 Material and Proposed Method
4.2.1 Method Framework
4.2.2 Data and Questionnaire
4.2.3 Attribute Characteristics
4.2.4 Data Mining Platform
4.2.5 Improved SOM Clustering
4.2.6 K-means Clustering
4.2.7 Improved K-means Clustering
4.3 Experimental Results
4.3.1 Improved K-means Assessment
4.3.2 Improved SOM Assessment
4.3.3 Final Projection Plot
4.4 Discussion
4.4.1 Why Only SOM and K-means
4.4.2 Comparison of Proposed Approaches
4.4.3 Final Comments
4.5 Summary
Chapter 5 An Epidemic Forecast Model for Diabetes Prevalence by Data Mining
5.1 Introduction
5.1.1 Problem Statement
5.1.2 Shortcomings of Existing Methods
5.1.3 Proposed Method and Significances
5.2 Material and Proposed Method
5.2.1 Model Strategy and Implications
5.2.2 Data and Features
5.2.3 Improved Clustering Analysis
5.2.4 Improved J48 Decision Tree Algorithm
5.3 Experimental Results
5.3.1 Measurements and Confusion Matrix
5.3.2 Regression Forecast Assessment
5.3.3 Cost/Benefit Accuracy Classifications
5.4 Discussion
5.4.1 Comparison with Other Decision Tree Algorithms
5.4.2 Final Comments
5.5 Summary
Chapter 6 An Accurate Clinical Implication Assessment Model of Diabetes Prevalence for Primary Health Education
6.1 Introduction
6.1.1 Prevalence Statement
6.1.2 Shortcomings of Prevalence Methods
6.1.3 Proposed Approaches and Significance
6.2 Material and Methods
6.2.1 Data Collection and Explanation
6.2.2 Attributes Selection
6.2.3 Attribute Parameters
6.2.4 Methodological Analysis
6.2.5 Rules Classification
6.2.6 Kappa Statistics
6.2.7 Logistic Regression Forecasting
6.3 Experimental Results
6.3.1 Methodology Assessment
6.3.2 Rule Forecast Assessment
6.4 Discussion
6.4.1 Comparison with Other Data Mining Algorithms
6.4.2 Final Comments
6.5 Summary
Conclusion
参考文献
Scientific Research Tasks and Major Achievements during Doctoral Study
致谢
燕山大学;