Public health systems are under pressure from rising costs, workforce shortages, increasing demand and limited resources. Data and AI can help organizations use their resources more effectively, improve care decisions and deliver better outcomes.
But the next phase of health care AI will depend on more than what the technology can do. It will depend on whether health leaders, clinicians, regulators and patients can trust it.
That was the central theme of a webinar I moderated with Christian Hardahl, EMEA Health Care Industry Leader at SAS and Saad Rais, Senior Manager in Health Data Science at the Ontario Ministry of Health. What emerged was a candid, practical conversation about where health care organizations are on their AI journey and what it will take to move forward with confidence.
The real barrier is not technology
The challenge for health care organizations is not simply about adopting new technology. The real opportunity lies in strengthening data foundations and improving data quality and getting the basics right before AI can deliver meaningful impact.
It's not an AI problem. It's more of a system problem or a data quality problem. Christian Hardahl, EMEA Health Care Industry Leader, SAS
Across the globe, digital maturity varies enormously, not just between countries but also within the same health region. Fragmented data landscapes, a lack of interoperability, and workflows not designed for data-driven decision-making all slow progress. AI is being explored and adopted, but innovation happens in pockets and the journey from pilots to real-world deployment remains slow.
Hardahl also pointed to a broader readiness gap. At HIMSS Europe 2026, the newly released State of Healthcare AI and Digital Transformation report revealed that in the US, only 7% of organizations have a mature AI governance strategy in place and just 21% of leaders believe their data is being fully leveraged.
The implication is clear. Before organizations can scale AI, they need to get the fundamentals right, strengthening data foundations, improving interoperability and then integrating AI and analytics into everyday decision-making.
From “Can we build it?” to “Can we trust it?”
Even where AI is being successfully deployed, health care leaders are asking different questions about AI than they were just a few years ago.
We've shifted from asking 'Can we build it?' to 'Can we trust it?' Christian Hardahl, EMEA Health Care Industry Leader, SAS
This shift reflects a growing recognition that, for health leaders, AI's trustworthiness is critical. Trust, grounded in governance, transparency and data quality, is the foundation for scaling AI safely and effectively. It must be designed in from the start, not bolted on after deployment.
Christian outlined some simple but telling questions organizations should be able to answer:
- Can you identify all your AI algorithms and who is accountable for each?
- Can you test algorithms on updated data in a controlled, automated process?
- Have you assessed your models for bias and is that bias at an acceptable level?
- Do you know how your data was created and where it originates?
- Do you have a governance model and a quality assurance board in place?
The regulatory landscape is adding further urgency. With regulations such as the EU AI Act, the European Health Data Space (EHDS), GDPR and HIPAA shaping the environment, governance and auditability are no longer optional. They are strategic imperatives. C-level leaders must be accountable for AI.
Encouragingly, leading organizations are already taking this seriously. SAS participates in the Responsible and Ethical AI in Healthcare Lab (REAiHL), a collaboration between Erasmus Medical Center, Delft University of Technology and SAS, endorsed by the World Health Organization.
As expectations from regulators, clinicians and patients increase, health systems must ensure that trust is built into every stage of AI deployment, not an afterthought.
AI in action: Real results from Ontario
The Ontario Ministry of Health offers a compelling example of what trustworthy analytics and AI can look like at scale. Advanced analytics and AI are embedded across the system to support policy decisions, operational efficiency and health outcomes. All underpinned by robust data governance and quality controls.
One standout example is AI-driven image recognition applied to ultrasound billing validation. What had previously been a manual, time-intensive process was transformed:
What used to take four weeks manually can now be done in under 30 minutes with near-perfect accuracy. Saad Rais, Senior Manager, Health Data Science, Ontario Ministry of Health
This kind of efficiency gain goes well beyond speed. It frees up clinical and administrative capacity, reduces error rates and allows teams to focus on higher-value activities.
Forecasting is another critical area. Predictive modeling in Ontario is being used to:
- Forecast and surveil diseases such as COVID-19, influenza and RSV.
- Identify patients at risk of adverse outcomes and predict high-cost users.
- Anticipate service demand and capacity pressures.
- Support emergency management and public health response to outbreaks.
By identifying patients at risk earlier, we can improve outcomes and avoid unnecessary cost pressures on the system. Saad Rais, Senior Manager, Health Data Science, Ontario Ministry of Health
Ontario’s experience demonstrates what becomes possible when data quality, governance and analytical capability are treated as foundational investments rather than afterthoughts.
Emerging technologies: RAG, generative AI and digital twins
The conversation also explored several technologies that are rapidly moving from experimentation to practical application.
Retrieval-Augmented Generation (RAG) has emerged as a particularly promising approach for health care, precisely because it addresses the trust problem head-on.
Rather than relying on general-purpose models trained on external data, RAG draws on an organization’s own vetted, proprietary knowledge base to generate citation-backed, source-grounded responses. This eliminates the risk of hallucinations, ensures outputs are traceable and auditable, and means sensitive data stays internal.
In Ontario, generative AI is already being used to interpret unstructured clinical notes from lab technologists, turning previously inaccessible data into actionable insights.
Digital twins are also generating growing interest. SAS partnered with Epic Games to create powerful, photorealistic 3D visualizations that allow health care teams to model, test and optimize real-world scenarios.
In Denmark, SAS is working with the Central Sterile Service in the Capital Region to run a 1:1 digital twin of an entire sterile processing facility, enabling real-time operational optimization and “what-if” scenario planning, including routing optimization for autonomous guided vehicles.
Trust is the foundation and the future
As health care organizations continue to modernize, the direction of travel is clear: towards interoperable health systems that use trustworthy data and AI to enable more personalized, preventive and equitable care.
In the short term, that means scaling proven use cases, improving governance and closing the gap between innovation and deployment.
But getting there requires deliberate investment in the foundations. Human oversight and judgment remain essential throughout and governance enables organizations to scale innovation responsibly. The organizations moving forward most effectively are those prioritizing three things:
- Interoperability – breaking down data silos and ensuring information flows where it is needed
- Governance – building accountable, transparent structures for managing data and AI
- Trustworthy AI – designing explainability, auditability and ethical safeguards into every stage of deployment
Because when data and AI are used in the right way, they don’t just improve efficiency, they help create health systems that are more responsive, more sustainable and ultimately, more effective for the populations they serve.
