Dr Alwin Tan, MBBS, FRACS, EMBA (Melbourne Business School)

Senior Surgeon | Governance Leader | HealthTech Co-founder |
Harvard Medical School — AI in Healthcare |
Australian Institute of Company Directors — GAICD candidate |
University of Oxford — Sustainable Enterprise

Institute for Systems Integrity

Introduction

Across Australia, a structural shift is underway.

Patients are gaining faster access to pathology and imaging results — often before a clinician has reviewed them.

This reform is framed as progress:
greater transparency, greater empowerment, greater access.

That is only part of the story.

This is not simply an access reform.
It is a reconfiguration of how information moves through the healthcare system.

And when information pathways change, decision-making changes with them.

The Hidden Shift: From Controlled to Distributed Interpretation

Healthcare has traditionally operated on a controlled sequence:

Signal → Clinician interpretation → Patient communication → Action

This structure was not accidental.
It ensured that:

• data was interpreted before it was acted upon
• uncertainty was managed
• escalation aligned with clinical risk

The emerging model introduces a fundamentally different pathway:

Signal → Patient access → AI/independent interpretation → Clinician engagement

Interpretation is no longer controlled.
It is distributed.

And once interpretation is distributed,
so is decision-making.

What the Evidence Actually Shows

The evidence on patient access is clear — and nuanced.

Access improves engagement.
Patients are more involved, more informed, and more satisfied with their care.

But the same evidence also shows:

• increased anxiety when results are abnormal
• difficulty interpreting results without clinical context
• increased communication burden on clinicians

In other words:

👉 Access improves participation
👉 But redistributes interpretive burden

That burden does not disappear.
It moves — upstream to patients, and back downstream into clinical workflow.

The AI Layer: Confidence Without Context

Into this new pathway enters AI.

Patients are increasingly using AI tools to:

• interpret results
• generate explanations
• form preliminary conclusions

This creates a second shift.

Not just distributed interpretation,
but accelerated interpretation.

However, current evidence is clear:

AI does not reliably improve decision quality in consumer use.
It can produce coherent explanations without full clinical context.

Which introduces a critical risk:

👉 confidence without validation

AI does not replace uncertainty.
It can mask it.

Where the System Begins to Fail

When early access and AI interpretation combine, three predictable failure patterns emerge:

1. Signal Distortion

Data is encountered without context.
Meaning is constructed before validation.

2. Interpretive Divergence

Patients and clinicians may arrive at different conclusions from the same result.

The consultation no longer begins with shared understanding.
It begins with misalignment.

3. Escalation Misalignment

Perceived urgency (patient/AI) diverges from clinical urgency.

This leads to:

• unnecessary anxiety
• inappropriate escalation
• delayed attention to true risk

These are not communication problems.

They are failures of information design.

The Governance Gap

Current reforms address one problem:

👉 access to information

They do not address the next problem:

👉 how that information is interpreted before clinical engagement

Historically, healthcare systems have governed:

• how data is generated
• how it is stored
• how it is transmitted

They have not governed:

👉 how meaning is constructed in the gap between access and action

That gap is now the most critical control point in the system.

The Shift Required: Interpretation Governance

If this reform is to succeed, the system must evolve.

Not by restricting access —
but by governing interpretation.

This requires deliberate design:

1. Context Framing

Results must be accompanied by:

• expected ranges
• clinical significance
• known limitations

2. Escalation Guidance

Patients must be told clearly:

• when to wait
• when to seek review
• when urgent action is required

3. AI Governance

Clear boundaries must be established:

• what AI can support
• what it cannot determine
• where clinical authority remains

4. Workflow Redesign

Consultations must adapt to:

• pre-informed patients
• pre-formed interpretations
• variable levels of understanding

5. Sensitivity-Based Release

Not all results are equal.

Some require:

• delay
• clinician-first communication
• structured follow-up

Conclusion

Transparency is necessary.

But transparency alone is not safe.

Access to information does not guarantee understanding.
And understanding does not guarantee correct action.

Australia’s digital health reforms are directionally correct.

But without interpretation governance, they risk:

• increasing anxiety
• distorting clinical prioritisation
• fragmenting decision-making authority

The question is no longer:

👉 “Should patients see their results?”

The question is:

👉 “How does the system ensure that meaning is constructed safely before decisions are made?”

Because in healthcare,

data is only useful if truth survives the journey from signal to decision.

References (Harvard style)

Alomar, D. et al. (2024) ‘The Impact of Patient Access to Electronic Health Records on Health Care Engagement’, Journal of Medical Internet Research, 26, e56473.

Australian Government Department of Health (2026) Modernising My Health Record – Improved access to health information.

Clark, M. (2024) Chatbots in Health Care: Connecting Patients to Information. NCBI.

Steitz, B.D. et al. (2021) ‘Immediate Release of Test Results and Clinical Workflow’, JAMA Network Open, 4(10), e2129553.

Steitz, B.D. et al. (2023) ‘Patient Perspectives on Immediate Access to Test Results’, JAMA Network Open.

World Health Organization (2025) Ethics and governance of AI for health. Geneva: WHO.

University of Oxford (2026) Risks of AI chatbots in medical advice.