Skip to content

References: Responsible AI and Agentic Systems

  1. Explainable artificial intelligence - Wikipedia - Overview of XAI techniques (SHAP, LIME, feature importance, saliency maps) that make model predictions interpretable, directly underlying this chapter's explainable AI section and its Bayesian diagnostic reasoning example.

  2. Algorithmic bias - Wikipedia - Survey of how biased training data and design choices produce unfair model outcomes across protected groups, grounding this chapter's model bias section and its readmission-model mitigation worked example.

  3. Multi-agent system - Wikipedia - Explains how multiple autonomous, communicating agents coordinate to solve tasks no single agent handles well alone, the architecture behind this chapter's care-coordinator example of specialized query, guideline, and review agents.

  4. Speech and Language Processing (2nd Edition) - Daniel Jurafsky and James H. Martin - Prentice Hall - Jurafsky and Martin are credited with the standard, widely adopted pipeline explanation for named entity recognition, text classification, and sentiment analysis that this chapter applies to a clinical note.

  5. Artificial Intelligence: A Modern Approach (4th Edition) - Stuart Russell and Peter Norvig - Pearson - Russell and Norvig originated the PEAS framework and the reflex/goal-based/utility-based/learning agent taxonomy that this chapter's tool-using agent and multi-agent system sections build directly on.

  6. What is Explainable AI (XAI)? - IBM - Practitioner-oriented explainer on why and how machine learning predictions are made interpretable to humans, complementing this chapter's discussion of explainability as a required, not optional, clinical AI feature.

  7. Classification: ROC and AUC - Google Machine Learning Crash Course - Interactive tutorial on ROC curves and AUC, the exact model evaluation metric this chapter's readmission-prediction worked example uses to compare logistic regression, random forest, and graph neural network performance.

  8. AI Risk Management Framework - National Institute of Standards and Technology (NIST) - Federal voluntary framework for managing AI trustworthiness and risk, formalizing the organizational review and accountability structures this chapter defines as AI governance.

  9. Building Effective Agents - Anthropic - Practical guide distinguishing predefined workflows from autonomous agents and detailing tool-use design patterns, directly relevant to this chapter's tool-using agent and multi-agent clinical workflow worked examples.

  10. Named Entity Recognition - GeeksforGeeks - Tutorial covering NER methods with a working Python/spaCy example, illustrating the entity-extraction stage this chapter's clinical NLP pipeline example applies to a raw shortness-of-breath clinical note.