In emergency departments, clinicians rely on CT (computed topography) scans to rapidly identify life-threatening conditions such as intracranial haemorrhage, midline shift, and critical spinal injury. Decisions on whether a patient needs urgent neurosurgery, intensive monitoring, or safe discharge are time pressured and are critical to patient outcomes.
While artificial intelligence shows promise in supporting CT‑based assessment of brain injury, translating that promise into clinical practice remains deeply challenging.
Key Points
- Deakin PhD students are driving research that bridges AI, clinical practice, and human-centred system design in neurotrauma care.
- Close partnership with The Alfred Trauma Centre and the National Trauma Research Institute ensure the work addresses clinical challenges with real constraints.
- Neurosurgeons and trauma specialists are co-investigators in developing AI-powered assessment of CT scans.
- Traumatic brain injury (TBI) is one of the leading causes of death and long-term disability worldwide
A clinician-centred research program
One strand of the program focuses explicitly on developing new methodological frameworks for clinician‑engaged AI research where access, ethics approvals, institutional constraints, and time limitations shape what is realistically possible.
Rather than assuming ideal conditions, the work documents and formalises:
- Asynchronous clinician input
- Iterative, non‑linear feedback loops
- Cross‑disciplinary translation between medical and AI literatures
- Adaptation to real‑world blockers, including ethics processes and institutional research agreements
This work establishes a replicable methodology that future researchers can use when working in similarly constrained clinical environments.
Progress across the program is being driven by Deakin PhD students Rin and Nguyet, who are working under the supervision of Associate Professor Rena Logothetis, at the intersection of AI, clinical practice, and human-centred design.
They are helping build clinically meaningful datasets, evaluate emerging AI models, and work with clinicians to design tools that fit real trauma workflows.
“Our PhD students are central to this work. They are developing the technical depth, clinical awareness, and ethical grounding required to translate AI research into tools that clinicians can trust.” – A/Prof Rena Logothetis
Building the data foundations AI needs
In the space of clinical use of AI, system‑building and human‑centred design are complimentary disciplines.
AI development focuses on building, testing, and performance. Human-computer interaction contributes theoretical and practical insight into usability, trust, and workflow integration. Neither is sufficient on its own.
In high‑stakes clinical environments, AI systems must not only work as expected, they must work with clinicians, respecting cognitive load, time pressure and decision accountability.
Achieving clinical consensus
Another major strand of the research centres on clinical consensus building.
Delphi methods allow structured consensus to emerge without requiring clinicians to be physically co‑located or synchronously available, making them particularly well suited to trauma research contexts.
The team is investigating:
- Clinician‑centred data annotation and standards development.
- Evaluation of emerging AI approaches for CT‑based brain injury assessment.
- Development of human-computer interaction‑informed models.
- Translation of fragmented real‑world practice into defensible research methods.
The research program is delivering immediate research outputs, and also establishing long-term national capability in safe, clinically-grounded AI.
Share