The pandemic accelerated telehealth adoption massively. Permanent MBS telehealth items followed. Telehealth is now a standard part of Australian healthcare delivery.

Two years into this new normal, we’ve learned lessons about integrating AI with telehealth. Some AI applications work well in telehealth contexts; others don’t translate from in-person care. Here’s what experience has shown.

The Telehealth-AI Intersection

Telehealth creates both opportunities and constraints for AI:

Opportunities:

• Digital-native environment where AI integrates naturally
• Recorded sessions enable AI analysis and documentation
• Remote data capture from connected devices
• Asynchronous options for AI-augmented communication
• Scale efficiencies across distributed patients

Constraints:

• Limited physical examination capability affects some AI applications
• Technology access and digital literacy barriers
• Network and technical issues affecting reliability
• Different patient-clinician dynamics than in-person care
• Regulatory and reimbursement frameworks still evolving

The AI applications that succeed in telehealth leverage the opportunities while working within the constraints.

What’s Working Well

Ambient Documentation

AI that records and transcribes telehealth consultations works particularly well:

• Recording is already happening (often required for clinical records)
• Adding transcription and documentation generation is natural extension
• Clinicians can focus on the patient rather than notes
• Documentation is generated from the actual conversation

Microsoft’s DAX, Nuance, and competitors are seeing significant telehealth uptake. The fit is better than for in-person consultations where recording is less established.

Triage and Symptom Assessment

AI-powered triage before telehealth consultations:

• Patients describe symptoms to AI before clinician connection
• AI generates preliminary assessment for clinician review
• Clinician starts with structured information rather than blank slate
• Consultation time focuses on clinical decision-making

This works better in telehealth than in-person because the digital interaction is already established.

Remote Patient Monitoring Integration

AI that analyses data from home monitoring devices:

• Vital signs, glucose, weight, and other measures captured at home
• AI identifies concerning trends or values
• Telehealth consultation focuses on AI-identified issues
• Clinician time is directed to patients who need attention

The combination of remote monitoring and telehealth with AI triage creates efficient chronic disease management models.

Post-Consultation Summaries

AI that generates patient-facing summaries after telehealth consultations:

• Plain-language summary of what was discussed
• Action items and next steps clearly stated
• Information the patient can refer back to
• Sent to patient automatically after consultation

Patients often struggle to remember consultation content; written summaries help. Generating these automatically saves clinician time.

Language and Accessibility

AI-powered real-time translation and captioning:

• Patients who speak languages other than English can participate more fully
• Hearing-impaired patients benefit from live captions
• Broader access to telehealth services

These accessibility applications are particularly valuable in telehealth where family members who might normally translate are less available.

What’s More Challenging

Diagnostic AI Requiring Physical Signs

AI that analyses physical examination findings doesn’t translate directly to telehealth:

• Skin lesion AI designed for close examination
• Auscultation AI requiring stethoscope input
• Movement analysis AI designed for in-person observation

Some workarounds exist (smartphone cameras, digital stethoscopes, home sensors) but they’re less reliable than in-person examination.

Complex Care Coordination

AI for complex care coordination often assumes in-person interaction:

• Multidisciplinary team meetings
• Family conferences
• Complex case discussions

While these can happen via telehealth, the dynamics are different. AI designed for in-person coordination may need adaptation.

Acute Care Applications

Most acute care AI is designed for emergency department or inpatient settings:

• Sepsis detection
• Deterioration prediction
• Critical result alerting

These don’t apply directly to telehealth (which is mostly non-acute), though some adaptation for virtual urgent care is emerging.

Implementation Lessons

From organisations that have integrated AI with telehealth:

Start With Documentation

Ambient documentation is the most mature, most accepted AI application in telehealth. It’s a sensible starting point:

• Clear value proposition (time savings)
• Established technology
• Natural fit with telehealth recording
• Relatively straightforward implementation

Build confidence with documentation AI before attempting more complex applications.

Ensure Reliability

Telehealth already has technical complexity. Adding AI that introduces additional failure points frustrates clinicians and patients.

AI tools must be:

• Reliable (minimal downtime)
• Fast (not adding latency to consultations)
• Degradable (telehealth works even if AI fails)

Pilot thoroughly before broad deployment.

Consider the Full Patient Journey

Telehealth isn’t just the video call. Consider AI across the full journey:

• Pre-consultation triage and preparation
• During-consultation support
• Post-consultation summary and follow-up
• Between-consultation monitoring

AI opportunities exist at each stage.

Address Privacy Clearly

Telehealth AI often involves recording, transcription, and data processing. Patients should understand:

• What’s being recorded
• How AI is processing their information
• How data is stored and protected
• Their rights regarding recordings

Clear communication builds trust.

Integrate With Existing Workflows

Telehealth platforms vary. AI needs to integrate with the platforms clinicians actually use:

• Video consultation platforms
• Practice management systems
• Electronic medical records
• Patient communication systems

Standalone AI that doesn’t integrate creates workflow friction.

For organisations navigating telehealth AI options, external expertise helps. AI consultants Melbourne and telehealth specialists can advise on integration approaches, though clinical workflow understanding is essential.

Reimbursement Considerations

Current MBS telehealth items don’t specifically address AI:

• Consultations are reimbursed; AI tools are not
• AI costs are borne by practices/organisations
• No AI-specific item numbers exist

This affects business cases for telehealth AI. Investment must be justified through:

• Efficiency gains (more patients per session)
• Quality improvements (better outcomes, fewer follow-ups)
• Clinician retention (reduced burnout, better work experience)
• Competitive positioning (attracting patients and clinicians)

Until reimbursement evolves, these indirect benefits must justify investment.

Regional and Rural Considerations

Telehealth has particular value for regional and rural patients who face distance barriers. AI integration should consider:

Connectivity limitations. Regional broadband may not support video and AI simultaneously. AI that works with lower bandwidth matters. AI consultants Brisbane have developed specific recommendations for regional telehealth AI that account for these infrastructure realities.

Access equity. AI tools should be usable by patients with limited digital access or skills. Don’t create further barriers for already-underserved populations.

Local context. Regional healthcare has different patterns and needs. AI trained on metropolitan data may not perform well regionally.

Indigenous health. Culturally appropriate telehealth and AI for Aboriginal and Torres Strait Islander communities requires specific attention.

Looking Forward

The telehealth-AI intersection will develop significantly:

Short term: Documentation AI becomes standard for telehealth-heavy practices. Remote monitoring integration matures. Triage AI grows.

Medium term: AI-augmented virtual care models emerge, with AI handling routine interactions and escalating to clinicians as needed. New care models become possible.

Longer term: The distinction between telehealth and AI-supported care blurs. All care involves some combination of in-person, remote, and AI elements.

Telehealth created infrastructure for AI. The two are increasingly inseparable.

For clinicians and organisations, the question isn’t whether to use AI in telehealth, but how to use it well—enhancing care while maintaining the human connection that makes healthcare meaningful.

---

Dr. Rebecca Liu is a health informatics specialist and former Chief Clinical Information Officer. She advises healthcare organisations on clinical AI strategy and implementation.