Three quarters of health system executives say the current delivery model is unsustainable without major change. Nine in ten are prioritizing digital and AI to fix it. The gap between acknowledging the problem and actually solving it has never been more expensive.
Healthcare has always been an industry where change moves slowly and the reasons are understandable. The stakes are high. The regulatory environment is complex. The people delivering care are already under significant strain, and adding new technology to an already pressurized environment without careful implementation can make things worse before they get better. These are legitimate reasons for caution. They are not legitimate reasons for the pace at which most health systems are actually moving in 2025.
A Chartis survey of 150 health system executives published in late 2025 captures the situation with unusual clarity. Three quarters of respondents said the overall sustainability of the current healthcare delivery model will not improve without significant structural changes. Nine in ten said they are prioritizing digital and AI capabilities to achieve those changes. And yet, across most health systems, the pace of meaningful digital implementation remains far below what the urgency of the situation warrants.
The disconnect is not primarily about technology. The best digital health technologies available today are mature, proven in comparable environments, and commercially available. The disconnect is about everything that has to happen inside an organization before technology can deliver on its promise — strategy alignment, data readiness, change management, and a willingness to redesign workflows rather than digitize the existing broken ones.
The healthcare systems delivering measurable outcomes from digital transformation share one characteristic: they redesigned care workflows before implementing technology, not after.
The Three Forces Making Digital Transformation Unavoidable
For health systems that have been watching digital transformation from a distance — running pilots, attending conferences, waiting for clearer evidence before committing — 2025 has introduced three pressures that are making that position increasingly difficult to sustain.
The first is workforce. The clinical workforce shortage that emerged from the pandemic has not resolved — it has deepened and broadened. Nursing shortages, physician burnout, and an aging workforce are reducing the human capacity available to deliver care using traditional models. Health systems that are not using technology to extend the productivity of their existing clinical staff — through AI-assisted documentation, automated administrative workflows, and intelligent scheduling — are finding themselves unable to see the volume of patients their communities need, not because of bed capacity but because of people capacity.
The second is data volume. Healthcare generates approximately 30 percent of the world's data, driven by imaging, diagnostics, wearables, and electronic health records. The organizations that can turn this data into actionable clinical and operational intelligence — identifying deteriorating patients before they crash, predicting readmission risk, optimizing staffing against demand — are delivering measurably better outcomes at lower cost. The organizations that cannot are sitting on the same data and getting none of the value from it.
The third is patient expectation. Patients — especially the under-55 demographic that will constitute an increasing share of healthcare demand over the next decade — expect the same digital convenience from healthcare that they get from banking, retail, and travel. Appointment booking, test results, care coordination, and billing that require phone calls and paper forms are not acceptable to this demographic. Health systems that do not offer a genuinely digital patient experience are already losing patients to those that do, particularly in elective and outpatient care where the relationship is less sticky than in primary care.
Why Interoperability Is Finally Starting to Matter
One of the most significant structural shifts in US healthcare's digital landscape in 2025 is the movement of interoperability standards — specifically FHIR R4 and R5, USCDI+, and TEFCA — from encouragement to enforcement. For years, electronic health records sat in silos. A patient's data at one health system was effectively invisible to providers at another, making coordinated care across organizations enormously difficult and creating significant duplication of effort and cost.
As interoperability requirements move into enforcement, health systems are being required to make patient data accessible in ways that create genuine opportunities for connected care. The organizations that are building data infrastructure with interoperability as a design requirement — rather than retrofitting it later — are positioning themselves to participate in care coordination models, value-based payment arrangements, and AI-powered population health programs that will increasingly define the competitive landscape in healthcare over the next five years.
AI-powered clinical decision support and predictive analytics are delivering measurable improvements in patient outcomes and operational efficiency — but only in organizations where data quality and governance are already strong.
What's Actually Working: AI Use Cases With Proven ROI
Amid the noise around digital transformation in healthcare, there are specific AI applications that have moved from promising to proven — delivering measurable improvements in clinical outcomes, operational efficiency, or both, in health systems at scale.
AI-assisted clinical documentation is the most broadly adopted and the one with the clearest return on physician time. Tools that listen to patient-physician conversations and automatically generate structured clinical notes are reducing documentation time by 30 to 40 percent in documented deployments. For physicians spending more than two hours per day on documentation — a figure that contributes significantly to burnout — this recovery of time is both operationally valuable and clinically significant, because it returns attention to patients rather than keyboards.
Predictive deterioration models — AI systems that monitor patient vitals and clinical data streams to identify patients at risk of sepsis, cardiac events, or other rapid deterioration — have demonstrated meaningful reductions in ICU transfers and mortality rates in peer-reviewed studies across multiple health systems. The technology works. The implementation challenge is integrating these systems into clinical workflows in ways that generate alerts clinicians trust and act on, rather than adding to the alert fatigue problem that already undermines patient safety in many settings.
Operational AI — demand forecasting, staffing optimization, surgical scheduling, and supply chain management — is delivering cost reductions that are significant enough to fund investment in clinical AI. A health system that can accurately predict emergency department volume 72 hours in advance can staff appropriately, reduce overtime, improve patient flow, and recover costs that currently disappear into unplanned overtime and agency staffing premiums. These are not transformational use cases in the clinical sense. They are financially material, and they create the organizational confidence and budget capacity to pursue more ambitious digital investments.
The Mistakes That Are Slowing Most Health Systems Down
The health systems that are struggling with digital transformation are not struggling because the technology doesn't work. In almost every case, the obstacles are organizational rather than technical.
The most common mistake is digitizing existing processes rather than redesigning them. A paper prior authorization process that becomes a digital prior authorization process with the same steps, the same approval chains, and the same turnaround time has not been transformed. It has been digitized. The workload has shifted from paper to screens but the underlying inefficiency is intact. Genuine transformation requires a willingness to ask which steps in a process actually add value and which exist because of historical inertia — and to redesign around the former rather than automating the latter.
The second mistake is underinvesting in change management. Technology implementation in clinical environments is hard because it requires changing deeply ingrained workflows among professionals who are already under significant strain and who have, in many cases, seen previous technology implementations create more problems than they solved. Clinician buy-in is not a soft consideration — it is a prerequisite for adoption. Health systems that involve clinical staff in implementation design from the beginning, that provide genuine training rather than checkbox compliance courses, and that create feedback loops so that problems get surfaced and fixed quickly, have measurably better adoption outcomes than those that do not.
The third mistake is treating cybersecurity as an afterthought. Healthcare is the most frequently targeted sector for ransomware attacks — the combination of sensitive data, operational criticality, and historically under-resourced IT security makes it an attractive target. A digital transformation that increases connectivity and data access without commensurately strengthening security posture is increasing risk faster than it is creating value. Every digital initiative in healthcare needs a concurrent security assessment, not a retrospective one.
The Vendors and Technologies Worth Evaluating Right Now
The healthcare technology market is large, noisy, and full of vendors making claims that are difficult to verify without deep domain expertise. For health system executives evaluating technology investments, a few categories deserve serious attention in 2025 based on both maturity and demonstrated impact.
Ambient clinical intelligence platforms — those that handle AI-assisted documentation and clinical note generation — have reached a level of maturity where the evidence base is strong and the implementation pathway is well understood. The leading platforms in this category have deployed at scale in major health systems and have published outcomes data. The question for most health systems is not whether to adopt but which platform fits their existing EHR infrastructure and clinical workflow context.
Healthcare-specific large language models and clinical decision support tools are maturing rapidly. The important distinction for procurement teams is between general-purpose AI tools applied to healthcare settings and tools that have been trained on clinical data, validated against clinical outcomes, and built with the regulatory requirements of a clinical environment in mind. The latter category is smaller but significantly more appropriate for high-stakes clinical decisions.
Patient engagement and virtual care platforms have evolved substantially from the telemedicine tools that were rapidly deployed during the pandemic. The current generation integrates with EHR systems, supports asynchronous care models for appropriate conditions, and provides the kind of consumer-grade user experience that drives adoption among patients who have choices about where they receive care.
What the Leading Health Systems Have in Common
The health systems consistently cited as leaders in digital transformation share characteristics that are more organizational than technological. They have executive leadership — specifically CMO and CIO alignment — that treats digital transformation as a clinical strategy, not an IT project. They have invested in data governance as a foundational capability before deploying AI on top of it. They have clinical champions — physicians and nurses who are genuine advocates for specific digital tools — embedded in the implementation process from design through deployment. And they measure outcomes at the patient and workflow level, not just at the technology adoption level.
None of these characteristics require exceptional resources or a large organization. They require a specific kind of organizational intention — a decision that digital capability is a clinical priority, not a technology project — and the willingness to sustain investment and attention through the implementation challenges that will inevitably arise. Health systems that have made that decision are increasingly visible in their communities and in the data. The ones still waiting are not waiting because they lack the technology options. They are waiting because the decision has not yet been made at the level of organizational conviction required to actually execute it.