Decision Support in Breast Cancer: Recent Advances in Prognostic and Predictive Techniques
The following sections are included:
Clinical Requirements for Decision Support Strategies in Breast Cancer
Risk-Group Assessment
Tumor-Biological Factors: The Plasminogen Activation System
The Role of Statistical Models in Clinical Decision Support for Breast Cancer
Intelligent Systems and Statistical Techniques for Prognosis and Prediction
Advanced Statistical Models as Intelligent System
The Cox Proportional Hazards Model and Generalizations
Classification and Regression Trees
Intelligent Systems as Statistical Models
Breast Cancer as a Complex Disease
A Representative Study
Node-Negative Breast Cancer: A Multivariate Proportional Hazards (Cox) Prognostic Model
Univariate Risk Assessment in Subgroups
Analysis of Time-Varying Effects
Classification and Regression Trees (CART)
Neural Survival Models
Neural Nets and Prognosis in a Clinical Setting
Previous Approaches to Survival Modeling Using Neural Nets
Neural Topology Used for Survival Modeling
Transformations and Input Neurons
Hidden Neurons
Output Nodes
A Survival Model Using Neural Nets to Define the Time-Dependent Risk Structure
Net Optimization
Initialization
Training
Complexity Reduction
Simulated Follow-Up Data
Simulation Technique
Simulated “Nonlinear Diseases”
Statistical Tools
Performance
Conclusions and Outlook
Acknowledgments
References
