Research that keeps health data where it belongs.
The research arm of Trenthos builds privacy-preserving technology to improve how healthcare is delivered in Australia - with patient consent at the centre, and data that never leaves the country.
Areas of focus, not finished work.
Much of the data and AI shaping healthcare was built on overseas populations. These are the directions we are building toward - research themes, honestly described, not capabilities already delivered.
Privacy-preserving machine learning
AI that learns from clinical data without that data ever being exposed or leaving its source.
Synthetic data for medical AI
Realistic, fully synthetic clinical data that supports research and software testing without real patient records.
Federated learning
Shared models trained across multiple sites, so everyone gets better tools while each site's data stays on its own premises.
Real-world evidence
Routine, consented, de-identified clinical information turned into insight that can improve care.
Clinical decision support
Tools that help clinicians make faster, better-informed decisions at the point of care.
Ophthalmic imaging analysis
How OCT, fundus photography and visual fields might surface useful signal - supporting clinical judgement, never replacing it.
Principles before products.
These commitments come first, and they are written to match the consent and privacy documents that govern how data is actually handled.
- Data stays in Australia Records and chronologies are stored and processed in Australia.
- Nothing without consent Patients decide whether their de-identified information is ever used for research, and can change their mind at any time.
- De-identified by default Research is designed around de-identified data with strong privacy protections.
- Clinics are partners Clinics that take part share in the outcomes - better tools, and co-authorship on research they help make possible.
- Transparency We explain what we do in plain language, and publish our methods openly where we can.
What we're learning as we build.
The technical and regulatory realities of building healthcare technology in Australia - what works, what is harder than it looks.
-
23 Oct 2026
For practices
Guardrails for AI in a medico-legal file: what the software must never do, and where each rule has to live
A guardrail is only as strong as the place it is enforced. Opinion drafting, page-checked facts, conflicts left unresolved, the brief as untrusted input and no training on case content, sorted by where each rule has to live.
-
14 Oct 2026
For practices
A medico-legal file under the Privacy Act: health information with a legal context layered on top
A brief is someone else's health history, collected from someone else, for one purpose. What the APPs ask of a practice, where software quietly sends files offshore, and how retention and breaches work.
-
6 Oct 2026
For clinicians
The allied health chronology: function, care and work across a medico-legal file
Occupational therapists and other allied health assessors read a claim file for what the person could do, who helped them and how that changed. The duties that apply, the care-hours threshold, the missing baseline, and where software fits.
A small team, building carefully.
Trenthos Research is founder-led and clinically informed. It is led by Dr Chun-Huei Liu, a clinically trained doctor building Trenthos and its products with a focus on responsible health technology in Australia. We would rather say less and mean it than overstate where we are.