Austin Health streamlines medical data access and reduces clinician burden

Austin Health, a major healthcare provider in Australia, faced mounting challenges from the growing complexity and volume of healthcare data. Clinicians struggled with significant cognitive burdens, requiring them to access data from multiple sources, slowing down processes and reducing face-to-face patient time. Austin Health conducted trials using Microsoft generative AI technologies to read and index two decades of scanned medical records, including handwritten notes, and to improve accessibility and searchability of clinical guidelines vital to daily operations. The AI system leveraged GPT-4 Turbo with vision capabilities and GPT-3 models to accurately recognize handwritten terms and rapidly retrieve precise internal guidelines. A key focus for Austin Health has been ensuring ethical, secure, and responsible AI usage. Data privacy controls restricted external internet access for AI models, and strict organizational policies prevented health data from being used to train third-party models. The initiative prioritized building evidence around AI's value, with early trials demonstrating faster and more accurate information retrieval for clinicians and strong support from medical staff. While production rollout is on hold due to government policy, internal tests highlight a foundation for future healthcare transformation, promising a shift toward non-document-based, knowledge-driven systems with voice capabilities for medical record-keeping and search. Austin Health is also preparing to integrate data from platforms like Oracle Health and Microsoft Dynamics, aiming for holistic patient and operational insights. The project underscores the importance of privacy, governance, and trust in clinical AI, laying the groundwork for future advances that could fundamentally change healthcare delivery in Australia.