An independent evaluation by researchers at the University of Technology Sydney (UTS) and St Vincent’s Hospital Sydney has found that an AI-enabled clinical decision-support tool is associated with improved antibiotic prescribing practices across rural and remote hospitals in Western Australia.
The tool, Clinical Branches, also known as Antimicrobial Pathways (AMPS), provides clinicians with step-by-step prescribing guidance based on individual patient characteristics and local antimicrobial resistance patterns. It translates clinical guidelines into practical decision pathways, supporting evidence-based prescribing where specialist infectious disease expertise may be limited.
The evaluation examined AMPS implementation across 68 Western Australian Country Health Service sites between 2023 and 2025. Researchers analysed 3,917 antibiotic prescriptions from three adopter and five non-adopter hospitals, alongside platform usage data and interviews with 30 clinicians and stakeholders.
After adjustment, prescribing appropriateness was 20.8 percentage points higher at adopter hospitals, while guideline compliance was 21.8 percentage points higher. Pharmacists and senior prescribers were among the most frequent users, reporting improved confidence in prescribing decisions.
However, adoption remained inconsistent. Of the 68 participating sites, 49 demonstrated little or no sustained use, while only nine achieved strong or sustained adoption. Barriers included limited awareness, insufficient training, workforce turnover, accessibility challenges, and difficulties integrating the tool into existing workflows.
Despite these limitations, clinicians generally expressed confidence in the technology, suggesting that implementation barriers rather than clinical resistance constrained adoption.
Dr Sarah Wise, senior lecturer in Health Services Management at UTS, highlighted opportunities to extend AI-enabled decision support beyond antimicrobial prescribing.
"The exciting next step is to incorporate real-world patient outcomes into decision support, using AI to develop 'living guidelines' that learn from what happens to patients and become increasingly responsive to individual patient characteristics and local conditions," she explained.
However, researchers emphasised that further investigation is required to determine whether improved prescribing translates into better patient outcomes.
The findings underscore the importance of structured training, workflow integration, and ongoing implementation support in scaling AI-enabled clinical decision tools, particularly across underserved healthcare settings.
Click here for the original news story.
