Week 1: Introduction to AI in Healthcare¶

Section 1: Foundations | Learning Outcome 1¶

Theme: Demystifying AI: From hype to operational understanding

Core Question: "What is this?"

Learning Objectives¶

By the end of this session, students will be able to:

Distinguish between artificial intelligence, machine learning, and traditional software
Identify the major categories of AI applications in healthcare
Articulate why AI literacy matters for healthcare practitioners

Content¶

1.1 Defining the Landscape¶

What AI Is (and Isn't)

Separating marketing claims from technical reality
The AI hierarchy: AI → Machine Learning → Deep Learning → Generative AI
Rule-based systems vs. learning systems: different approaches, different strengths
Why "AI" has become an overloaded term

Why Healthcare is Both Promising and Challenging for AI

Rich data availability vs. data quality challenges
High-stakes decisions requiring reliability
Regulatory and ethical complexity
Workforce implications and resistance

1.2 Categories of Healthcare AI¶

Diagnostic Support - Medical imaging analysis (radiology, pathology, dermatology) - Differential diagnosis assistance - Early detection and screening applications

Clinical Decision Support Systems - Alert and reminder systems - Drug interaction checking - Guideline-based recommendations

Predictive Analytics - Patient deterioration prediction (e.g., sepsis, cardiac arrest) - Readmission risk scoring - Resource demand forecasting

Administrative AI - Scheduling optimisation - Clinical documentation and coding - Revenue cycle management

Conversational AI - Virtual health assistants - Symptom checkers - Mental health chatbots

1.3 The Australian Healthcare AI Context¶

Current State of Adoption - Where AI is actually deployed in Australian health services - Gap between pilot projects and operational systems - Barriers to adoption: technical, cultural, regulatory

Key Stakeholders - Clinicians: end users and domain experts - Health services: purchasers and implementers - Regulators: TGA, AHPRA, state health departments - Vendors: commercial providers and their incentives - Patients: ultimate beneficiaries and risk-bearers

Unique Challenges of Federated Australian Healthcare - State-based health systems - Public-private mix - Geographic dispersion and rural health - Indigenous health considerations

Aeromedical Thread¶

AI Applications in Retrieval Medicine: An Overview¶

The aeromedical context presents unique opportunities and challenges for AI:

Current and Emerging Applications - Triage and dispatch decision support - Resource allocation optimisation - Patient deterioration prediction during transport - Point-of-care diagnostic assistance - Clinical documentation and handover support

Unique Constraints - Remote decision-making with limited backup - Connectivity limitations during transport - Time-critical interventions - Small, high-performing teams - Cross-jurisdictional operations

Why This Context Matters Throughout this course, we'll use aeromedical retrieval as a lens for examining AI. The high-stakes, resource-constrained, time-critical nature of retrieval medicine amplifies both the potential benefits and risks of AI systems.

Learning Activities¶

Pre-Class Preparation¶

AI Literacy Self-Assessment
Complete the provided questionnaire to establish your baseline understanding
Identify your current AI encounters in practice
Media Analysis Exercise
Read the three provided media articles about healthcare AI
Note the specific claims made about AI capabilities
Identify any red flags or unsupported assertions

In-Class Activities¶

Marketing vs. Reality Analysis (Small Groups, 30 mins)
Groups receive AI product marketing materials
Task: Identify specific claims and categorise as verifiable, plausible, or hype
Share findings and discuss common patterns
AI Touchpoint Mapping (Pairs, 20 mins)
Map all current or potential AI touchpoints in your clinical environment
Categorise by: diagnostic, decision support, predictive, administrative
Identify which you interact with vs. which operate invisibly
Large Group Discussion (30 mins)
What surprised you about AI in your environment?
Where is AI most/least appropriate in healthcare?
What would you need to know before trusting an AI system?

Post-Class Activities¶

Reflective Journal Entry (300 words)
Describe a personal encounter with AI in clinical practice
What did you understand about how it worked?
How did you decide whether to trust its output?

Practical Exercise 1: Exploring AI Outputs (Colab)¶

📓 Open Exercise 1 Notebook

Objective¶

Interact with pre-trained AI models to observe their behaviour and limitations.

Exercise Overview¶

Using a provided Colab notebook, you will:

Part A: Run a Pre-trained Image Classifier - Load a model trained to identify chest X-ray findings - Submit sample images and observe predictions - Note confidence scores and how they vary with image quality

Part B: Interact with a Clinical Language Model - Use a provided API connection to query a medical LLM - Test it with clinical scenarios and observe responses - Document instances where the model is helpful vs. problematic

Part C: Compare Rule-based vs. ML Approaches - Run a traditional clinical decision rule (e.g., Wells criteria calculator) - Run an ML-based prediction on the same scenario - Reflect on differences in transparency and behaviour

Key Questions to Address¶

What did the AI get right? What did it get wrong?
How confident was the AI in its predictions? Was that confidence warranted?
What would happen if you relied on this AI in your clinical practice?

Deliverable¶

Short reflection (300 words) comparing AI behaviour to your clinical expectations.

Indicative Resources¶

Required Reading¶

Topol, E. (2019). Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. Chapter 1: Introduction.

Multimedia¶

[Video] What is Machine Learning? (Google, 10 mins) [Link to be added]
[Podcast] AI in Healthcare episode selection [Link to be added]

Session Summary¶

This week established the foundational vocabulary and conceptual framework for understanding AI in healthcare. Key takeaways:

AI is a broad term encompassing many different technologies with different capabilities
Healthcare AI spans diagnostic, predictive, decision support, and administrative applications
Australian healthcare presents unique opportunities and challenges for AI adoption
Aeromedical retrieval amplifies both the potential benefits and risks of AI
Critical evaluation of AI claims requires understanding what AI actually does

Next Week: We'll dive deeper into how AI systems work—the architecture and algorithms that power healthcare AI applications.