Modern healthcare decision-making relies on data. However, until now, fragmented systems and legacy technologies have compromised the quality of data that health professionals rely on to perform their roles. According to Lou Compagnone, Director of Artificial Intelligence at Datacom, AI can help organisations overcome these challenges–but only if they take a coordinated, concurrent approach to data modernisation and AI adoption.
Fragmentation impedes efficiency and increases risk
Data fragmentation is widely acknowledged as a severe problem in healthcare. In 2024, the Productivity Commission noted that making better use of data in electronic medical records systems could save billions of dollars per year.
According to Compagnone, clinical notes, administrative records, billing information and scheduling data is typically distributed across dozens of platforms, most of which were never built to talk to each other. This creates problems that extend well beyond IT. “Fragmentation isn’t just an IT problem–it’s a care delivery problem,” she says. “Practitioners move between wards, facilities and care settings throughout the day and need the right patient information to follow them. When it doesn’t, decisions slow down and risks go up.”
Fragmentation also heightens cyber security risk in an industry targeted heavily by adversaries who use ransomware and other malware to exfiltrate sensitive data or threaten to disrupt life-critical services in hospitals.
Fragmentation a key problem in healthcare
Healthcare organisations need to address these issues in a highly regulated environment that requires them to comply with myriad federal, state and territory laws and regulations covering sensitive issues such as patient data sovereignty, consent and compliance.
The rapid take-up of AI is adding to scrutiny of the sector; for example, healthcare practitioners must exercise diligence and care when using AI tools for treatment, with the Australian Health Practitioner Regulation Agency noting that patients must give informed consent to the inputting of their personal data into AI models and tools.
Treat data readiness as a strategy problem
To successfully overcome these challenges, says Compagnone, healthcare organisations need to treat data readiness as a strategy problem rather than a technology issue. “Before anyone talks about platforms or tools, they need to answer fundamental questions such as what data they hold, where it lives, how is it classified and how does it flow through the organisation day to day?” she says. “The answers to these questions should determine where and how data is located and managed.”
This approach should inform specific conversations about data location and safeguards for each AI use case, with security addressed as a foundational concern. “For any AI workload involving protected health information, the starting point should be ‘never assume, always verify’, with access controlled at the data level,” says Compagnone.
Start with internal processes to optimise AI
She points out that the safest path to AI value starts with workforce management, operations, documentation and data infrastructure, rather than patient-facing care. “Starting with back-office AI gives organisations a chance to figure out what works before taking on the complexity and risk exposure of clinical or patient-facing applications. Improved administrative workflows, smarter rostering, automated documentation, and better data access for non-clinical staff all create direct upstream benefits for the people delivering care.”
Deliver data modernisation and AI adoption concurrently
Rather than treat data modernisation and AI adoption as sequential steps, Compagnone says, healthcare organisations should work on them concurrently across two coordinated tracks. “Track One is the foundational work, which includes building or modernising a data platform, establishing data classification, governance and sovereignty guardrails and fixing legacy application debt,” she explains. “This gets your data to a place where it’s knowable, accessible and trusted.
“Track Two is about quick wins with guardrails. For example, creating a safe, AI-powered personal productivity tool for staff–a low-risk, high visibility win to build internal confidence and executive support.”
A clear vision essential to AI success
Overall, according to Compagnone, successful AI implementation includes a clear vision for the role AI will play in achieving business objectives, backed by a comprehensive strategy that addresses key pillars such as optimisation of business functions, technical foundations, data governance and talent development. It also entails working closely with clinical team members and other stakeholders to determine where AI adds value.
“There are three questions you should bring into your next leadership conversation,” says Compagnone. “How much data do we have and how much of it is usable? Where can we create value for our people right now? And ‘Do we have the governance in place to scale?’
“By establishing a robust foundation with resilience, connectivity and flexibility, healthcare organisations can answer these questions positively and execute an AI-powered transformation that improves patient care and operational efficiency.”
To explore how your healthcare organisation can build the right foundations for AI at scale, contact the Datacom team at enquiriesIP@datacom.com to start the conversation.
