You can’t build the future of healthcare research on data you cannot access

As part of our new “The future of life sciences R&D” series, exploring how emerging technologies – from AI and digital trials to synthetic biology and advanced data platforms – are reshaping the future of drug discovery, clinical research, and scientific innovation, we invited Alastair Robertson, Chief Information Officer at Resmed, to share his views.

Here, Alastair argues that healthcare’s greatest AI challenge is not building smarter models, but creating accessible, connected, and trustworthy data ecosystems. Drawing on developments in connected health, cloud infrastructure, and real-world patient data, Alastair explores why interoperable systems are the foundation for more predictive research, personalised care, and the next generation of healthcare innovation.

Artificial intelligence is rapidly reshaping healthcare. From accelerating research to identifying disease patterns and personalising treatment, the opportunities are significant. But amid the excitement surrounding increasingly sophisticated AI models, many organisations are overlooking a more fundamental challenge: the infrastructure needed to make healthcare data accessible, connected and actionable.

The real question for healthcare leaders is not simply what AI can do. It’s what value AI can actually create for patients, for clinicians, and for researchers. And right now the answer to that question is being limited by the foundations that are meant to support the intelligence layer.

Healthcare is generating and collecting more patient data than ever before. Connected devices, remote monitoring tools, wearables and digital platforms are creating a continuous stream of information about how people live, sleep, breathe, and manage their health.

The catch? Much of that data remains inaccessible, disconnected or unactionable.

As a result, healthcare’s AI challenge is not a lack of intelligence, it’s a lack of connectivity. AI can help address these inefficiencies, but only when it is built on an interoperable infrastructure that makes data accessible, usable, explainable and scalable.

This is the reality behind the AI hype in healthcare, and it’s why the future of healthcare research will be determined by the infrastructure that supports breakthroughs in AI.

Poor data foundations are undermining healthcare’s AI revolution

Over the last few years, most of the attention and investment has been focused on AI models and algorithms which promise to spot diseases sooner or personalise treatment to individuals’ needs. But these sophisticated tools rely on the assumption that the data fuelling the systems is available, organised and reliable.

Right now, it’s not.

Gartner estimates that 60% of AI projects will be abandoned without AI-ready data foundations in place. This reflects a consistent underinvestment in the foundational technology layers that make AI possible in the first place; cloud infrastructure, system interoperability, data governance, privacy frameworks and workflow integration.

In healthcare, data is the ticket to play. Privacy must prevail, but systems also need to talk to one another if data is going to become actionable. When connected responsibly, data from across the health journey can contribute to better healthcare for every patient, not just those who are already visible within the system.

The foundations that actually matter

Fixing this challenge starts with getting the infrastructure right, in three interconnected layers. 

First, secure cloud infrastructure that can handle the processing and storage of data at scale for millions of patients. Second, interoperability that ensures the systems which hold data are communicating with each other and are working together effectively. And third, governance that implements clear rules and guardrails around privacy, consent, security and data quality that protect patient information and give meaningful choice with how their data is used. 

Each layer depends on the other, and it’s what makes data usable and earns people’s trust. 

However, none of this is useful if the underlying data is incomplete or inconsistently structured. AI is only as effective as the data it is built on, and in healthcare, often large parts of a person’s health journey never make it into digital systems in the first place. 

Most of the story happens outside the clinic 

Traditionally research and care have been built around a visit to the clinic but this is only a snapshot of the patient’s life, rather than the full story.

In sleep health, the awareness gap is striking. Snoring, fatigue, poor concentration and disrupted sleep are routinely normalised as stress, ageing or the cost of modern life. Many people live for years with symptoms they do not recognise as signs of a treatable condition. Whilst 64% of Australians say they’d likely seek professional help for a sleep issue, only 22% actually have. Not only is the clinic getting a limited picture of each patient, it’s also only getting that picture for the minority of people. 

This is where AI can create a bridge between everyday curiosity and clinical conversation. Not by replacing clinicians or turning consumer devices into diagnostic tools, but by making health insights more accessible and more connected to care. Helping someone move from ‘I’m just tired’ to ‘maybe I should learn more about my sleep health’ is itself a meaningful intervention, and it depends on having the infrastructure to support that journey when it begins.

Meanwhile, people are already generating health data every day. Four in ten Australians now check their sleep data via a wearable device, up from 18% the year before. We’re making progress in how people actively track their health, but most of that data is not connected to pathways of care. It sits on a phone. Useful to the individual, yes. But it could be doing so much more.

This is a gap that better infrastructure can close. In practice this looks like devices that can continuously record a patient’s sleep throughout the night within their own home, syncing data to a secure cloud-connected care platform where it becomes visible to their clinical team in real time. This enables the clinicians to proactively intervene without any delays of waiting for the next appointment or scheduled visit.

Predicting future outcomes instead of documenting past ones

This shift in data infrastructure also has the power to change the nature of how research is conducted and what it is capable of seeing.

Machine learning can recognise patterns across large and complex datasets that are too subtle, distributed and unintuitive for humans to identify consistently on their own. When AI is built on well-structured, continuously updated data, it can detect those signals, and surface insights that would not otherwise be visible or actionable at all.

Traditional life sciences research is largely retrospective. A hypothesis is formed, a study is designed, participants are recruited, data is collected over months or years, and conclusions are drawn after the fact. The question is always ‘what happened?’, so the answer, when it arrives, is already historical.

But continuous, real-world data allows us to start asking “what is likely to happen next?” 

When a care platform is constantly receiving and understanding large amounts of data, researchers can observe how health evolves, identify early signals that precede deterioration, and develop predictive models based on real life experiences, not just how they present in a controlled clinical setting. 

In sleep and respiratory care, this could mean identifying changes in breathing patterns, sleep efficiency and device adherence trends that predict a problem weeks before it becomes clinically apparent. In this scenario, the research moves from documenting outcomes to anticipating them. 

But the real opportunity is not simply to predict deterioration, but to enable the right action at the right time. This also changes how hypotheses are formed and tested. Rather than designing a study around a rigid plan and waiting for it to go live, researchers can work iteratively, using live data to find patterns, form hypotheses and constantly improve and refine in real time. 

At the population level, predictive signals can redesign care models around early intervention rather than reactive response and treatments can move to ongoing adjustments based on how a patient is responding day to day.

Responsible healthcare AI must be judged by how clearly accountability is defined, how transparently recommendations are generated, how effectively privacy and consent are protected, and how safely insights are translated into action.

The infrastructure is the strategy

For health technology leaders trying to move from AI ambition to AI readiness, investing in the foundations that can support these models reliably, at scale and responsibly is the key.

Get that right, and the possibilities shift from reactive to proactive care, where treatment is shaped around individuals rather than around population averages, and from data generated outside the clinic as part of everyday life. The best AI in healthcare will make care more accessible earlier, more personalised, and more effective and the value it creates will be something patients can actually feel. 

You cannot build the future of healthcare research on data you cannot access. The right infrastructure to access, connect and act on that data is not just a technical requirement, it is the foundation on which the future of healthcare research and patient care will be built.