Research

Machine learning and diagnosis

Hilary Bowman-Smart, University of South Australia

Melissa McCradden, University of Adelaide

A formal diagnosis, as recorded for administrative or record-keeping purposes, is often critical to access medication, subsidised treatment or care services. A formal diagnosis requires someone with the appropriate epistemic authority to declare it, such as a medical doctor. However, the labour and expense associated with formal diagnosis is considerable, and there are significant inequities in access to diagnosis for rural and marginalised populations. Technologies like genomics and machine learning applications are being integrated into the diagnostic process as potential solutions but carry significant ethical challenges. Applications of machine learning includes diagnostic prediction from imaging data, decision-making supports in triage, and the use of large language models as part of the clinical reasoning process.

Through a qualitative interview study, this project will provide critical insight into the experiences of healthcare professionals with the changing landscape of diagnostic technologies and inform the ethical integration of machine learning applications into healthcare in Australia. In addition to an ethical analysis, this project will explore the epistemic authority of machine learning models and how they might affect the social, administrative and legal legitimacy of a diagnosis.