Chapter Contents:

• 7.1 Background
• 7.2 A generative model of annotators
• 7.2.1 The ground truth model
• 7.2.2 The annotator model
• 7.3 Bayesian probability in parameter estimation
• 7.4 The Bayesian continuous-valued label aggregator
• 7.4.1 The MAP approach of the BCLA model
• 7.4.2 Convergence criteria for the BCLA-MAP model
• 7.4.3 Learning from incomplete data using the BCLA-MAP model
• 7.5 Data description
• 7.5.1 Simulated QT dataset with independent annotators
• 7.5.2 The 2006 PhysioNet challenge QT dataset
• 7.5.3 Methodology of validation and comparison
• 7.6 Results and discussion
• 7.6.1 Simulated dataset
• 7.6.2 PCinC QT dataset
• 7.7 Conclusion and future work
• Acknowledgement
• References