Digital Health and Care Delivery: Market Readiness Across Health Systems

The argument in brief

The question facing health ministries, payers and investors is no longer whether care will be delivered partly through digital channels, but how quickly, how equitably, and on whose infrastructure. The pandemic period compressed a decade of behavioural change into two years: video consultations moved from the margins to routine, remote monitoring entered mainstream chronic-disease management, and clinical records finished going digital in most high-income systems. What followed was less a straight-line ascent than a settling — a partial retreat from peak usage, a reckoning with evidence, and a slow negotiation over who pays. This report assesses where the sector actually stands, stripped of promotional framing, and what would have to be true for digital care to scale durably by 2030.

Our central finding is that readiness is constrained less by the maturity of technology than by the surrounding conditions that determine whether a working tool becomes standard practice. Four constraints recur across every system we examined: whether there is a stable way to pay for digital care, whether data can move safely between systems, whether the workforce has the capacity and incentives to change how it works, and whether the public trusts the handling of health data. Systems that have addressed these conditions — a small group spanning parts of the Nordics, Israel, Singapore, Estonia and pockets of larger systems — are scaling. Systems that have not are accumulating pilots that never reach production, a pattern common enough to have earned the label "pilotitis" in the global-health literature.

On market structure, we treat digital health as six overlapping segments rather than one figure, and we are explicit that published totals vary by a factor of two or more depending on scope. Our bottom-up estimate places 2024 global revenue in the region of US$275–440bn, dominated by clinical information technology and growing fastest, in percentage terms, in analytics and clinical AI from a comparatively small base. These are estimates built on stated assumptions, not measured accounts, and we flag them as such throughout.

US$275–440bn

Central estimate, 2024 global digital-health revenue

bottom-up planning estimate, net of overlap

6

Overlapping segments, not a single market

US$9–10tn

Global health spending

on the order of a tenth of world output

≈10m

Projected health-worker shortfall by 2030

WHO; falls largely on low- and middle-income systems

Looking to 2030, the plausible range of outcomes is wide and turns on policy choices more than on invention. Reimbursement decisions, interoperability mandates, AI regulation and cybersecurity resilience will do more to shape the trajectory than any single product category. We set out three scenarios — durable integration, uneven acceleration, and stalled progress — and the near-term decisions that move a system between them.

Key takeaways

• Digital health is a portfolio, not a market. The six segments — clinical IT, virtual care, remote monitoring and connected devices, consumer mHealth and wearables, health analytics and AI, and digital therapeutics — behave differently on growth, margin and regulation. Aggregating them into a single "market size" obscures more than it reveals and invites double-counting.
• Payment is destiny. Across systems, the durability of virtual-care volumes tracks the durability of reimbursement. Where emergency payment parity was withdrawn or left uncertain, usage fell back toward a narrow set of use cases; where it was made permanent, usage stabilised at a structurally higher level.
• Interoperability is the rate-limiting step for value. The clinical and economic case for digital care depends on data moving between settings. The convergence on the HL7 FHIR standard is real progress, but semantic interoperability — shared meaning, not just shared format — remains partial almost everywhere.
• Leapfrogging is real but conditional. Middle-income systems investing in digital public infrastructure — unique identity, provider and facility registries, and a national health data exchange — can extend population-level services faster than wealthier systems weighed down by incompatible legacy estates. The conditionality is governance and trust, not technology.
• AI is crossing from back office to bedside slowly and unevenly. Ambient documentation and administrative automation are diffusing quickly because the risk is contained; diagnostic and decision-support tools face a higher bar of evidence, liability and monitoring, and adoption reflects that.
• Cybersecurity is now a clinical-safety issue. Ransomware against hospitals and health data intermediaries has demonstrated that digital dependence without resilience is a patient-safety exposure, not merely an IT cost.
• The digital divide is reproduced inside health. Connectivity, device access, language and digital literacy determine who benefits. Without deliberate design, digital care widens the very inequities health systems exist to narrow.

1. Context and why it matters

Health systems everywhere are being asked to do more with structurally constrained resources. Global health spending reached roughly US$9–10 trillion, on the order of a tenth of world output, yet the World Health Organization estimates that around half the world's population lacks full coverage of essential services and that a projected shortfall approaching ten million health workers will fall largely on low- and middle-income countries by 2030. Populations are ageing — the share of people aged sixty and over is rising steadily, and with it the burden of chronic, co-occurring conditions that require continuous rather than episodic care. These pressures are not cyclical; they are structural, and they explain why digital delivery has moved from an efficiency conversation to a capacity conversation.

Digital care is attractive because it promises to change the unit economics of three things at once: where care happens, who delivers it, and how continuously it can be monitored. Moving a routine follow-up from a clinic to a video call, shifting stable chronic-disease surveillance from periodic visits to continuous remote monitoring, and automating documentation so clinicians spend more time with patients — each targets a specific bottleneck in a workforce-constrained system. The difficulty is that these gains are real only under conditions that many systems have not yet built: connectivity, interoperable records, a payment mechanism, trained staff and public confidence.

It is worth being precise about what changed during the pandemic period and what did not. Behaviour and regulation changed abruptly; underlying infrastructure and financing did not. Temporary emergency measures — payment parity for virtual visits, relaxed licensing across jurisdictions, waivers on prescribing — created a demand shock without permanently changing the plumbing. The years since have been a test of which changes would be made structural. That test is still running, and its outcome is the central uncertainty this report addresses.

1. Pre-2020
   Baseline

   Video consultations at the margins; clinical records still finishing the move to digital in most high-income systems.

2. Pandemic period
   Demand shock

   A decade of behavioural change compressed into two years — emergency payment parity, relaxed cross-jurisdiction licensing and prescribing waivers created a demand shock without changing the plumbing.

3. Post-peak (now)
   Settling and reckoning

   A partial retreat from peak usage, a reckoning with evidence, and a slow negotiation over who pays.

4. To 2030
   Three scenarios

   Durable integration, uneven acceleration, or stalled progress — decided by policy choices more than by invention.

2. Market structure and scale

We define digital health as the application of information and communications technology to the financing, management and delivery of care, spanning clinical, consumer and infrastructural layers. Published estimates of "the digital health market" range from roughly US$200bn to well above US$600bn, and the spread is almost entirely a function of scope: whether health IT infrastructure, consumer wearables, life-sciences analytics and telecom-adjacent services are included. Rather than defend a single number, we disaggregate into six segments and give transparent ranges.

The estimates below are a bottom-up synthesis built from public company disclosures, national health-IT expenditure data, multilateral connectivity statistics and analyst syntheses, reconciled to avoid the most obvious double-counting (for example, remote-monitoring hardware sold into telehealth programmes). They are planning estimates, not measured accounts, and the ranges are deliberately wide.

• Clinical IT45%
• Virtual care19%
• Consumer mHealth & wearables15%
• Remote monitoring & devices12%
• Health analytics & clinical AI7%
• Digital therapeutics2%

Three structural features of this market matter for readiness. First, the largest segment is the slowest-growing and the least contestable. Clinical IT is dominated by a small number of entrenched vendors with high switching costs; it is where most digital-health spending already sits, and it is where incumbency, not innovation, sets the pace. Second, the fastest-growing segments are the least proven on unit economics. Analytics, AI and digital therapeutics show the steepest percentage curves precisely because they start small and because their durable margins and clinical value are not yet settled. Third, consumer and clinical layers are converging but not merged. Wearable data is increasingly clinically relevant, yet the pathways to bring consumer-generated data into regulated clinical workflows — and to pay for the clinician time it generates — remain immature.

For decision-makers, the implication is that "market size" is the wrong lens. The relevant questions are segment-specific: in clinical IT, the question is interoperability and vendor concentration; in virtual care, reimbursement durability; in AI, evidence and regulation; in consumer health, the bridge into clinical use.

3. Drivers and adoption dynamics

Four demand-side forces are pushing digital adoption, and they are unusually well-aligned. Demographics and chronic disease create sustained demand for continuous, lower-cost monitoring rather than episodic visits. Workforce scarcity makes any tool that extends clinician reach — asynchronous consults, automated documentation, triage support — structurally attractive rather than merely convenient. Cost pressure on payers, public and private, rewards care that can be delivered at lower marginal cost or that prevents expensive downstream episodes. And connectivity has expanded to the point where roughly two-thirds of the world's population uses the internet, with mobile networks reaching considerably further — though the remaining unconnected population is disproportionately rural, poor and in exactly the settings where health need is greatest.

Against these drivers sit equally important frictions that determine whether demand converts into durable practice. The most decisive is reimbursement. The natural experiment of the pandemic period is instructive: where emergency payment parity for virtual care was allowed to expire or left in policy limbo, usage reverted toward a narrow band of behavioural-health and specialist follow-up cases; where payment was made permanent and predictable, usage settled at a structurally higher plateau. Providers respond to how they are paid, and no amount of clinical enthusiasm compensates for an absent billing pathway.

The second friction is interoperability. The value of digital care compounds when data follows the patient across settings; it collapses when each system is a silo. The industry has largely converged on the HL7 FHIR standard for exchanging data, which is genuine progress at the level of format. But semantic interoperability — ensuring that a coded diagnosis, medication or result means the same thing across systems — remains partial. Interoperability is as much a governance problem (who is permitted to access what, under which consent model) as a technical one.

The third is workforce capacity and trust in tools. Digitisation redistributes work before it reduces it; poorly designed systems have been a documented contributor to clinician administrative burden and burnout. Adoption succeeds where tools demonstrably remove work — ambient documentation is diffusing quickly for precisely this reason — and stalls where they add clicks. The fourth is public trust in data handling, which is fragile, easily damaged by breaches or perceived commercial overreach, and slow to rebuild. Trust is not a soft factor; in several systems, data-sharing initiatives have been delayed or withdrawn after public confidence eroded.

4. A readiness framework and comparative lens

Because the constraints are consistent, readiness can be assessed against a common set of dimensions. We use six: connectivity and infrastructure, data governance and interoperability, financing and reimbursement, workforce digital capacity, regulation and safety oversight, and public trust. No system scores uniformly high; the useful question is which dimension is each system's binding constraint.

Grouping systems by their dominant pattern yields four archetypes. These are analytical simplifications — most real systems straddle categories — but they clarify where different systems can and cannot move quickly.

The Integrators demonstrate what a mature system looks like: near-universal digital records, e-prescribing as the default, a national health data exchange, and — critically — a public that broadly accepts data sharing under a clear consent and transparency regime. Estonia's long-standing e-health infrastructure and the Nordic systems are the reference cases. Their remaining work is incremental: extending analytics and AI on top of a foundation that already exists.

The Scalers, counterintuitively, include the wealthiest systems. The United States has near-universal EHR adoption in hospitals and high adoption among office-based physicians, world-leading investment, and a deep vendor ecosystem — yet fragmentation across payers and provider networks, and a payment system of great complexity, mean that data liquidity and consistent reimbursement lag the level of investment. Germany combines strong capacity with a cautious, consent-heavy data culture. For Scalers, the constraint is connective and financial, not technological.

The Leapfroggers are the most consequential story of the decade. Rather than digitise institution by institution, several middle-income systems are building shared digital public infrastructure — verifiable identity, provider and facility registries, and health-data exchange — as a public good on which services are layered. India's national digital health programme, anchored on health identifiers and an exchange architecture, is the largest such effort; Indonesia and Rwanda are pursuing comparable stacks. These systems can, in principle, reach population scale on coverage faster than Scalers because they are not unwinding legacy fragmentation. Their risks are governance, workforce readiness and maintaining public trust while operating at extraordinary scale.

The Early-stage systems face the foundational constraints — intermittent connectivity, limited device access, thin infrastructure and donor-dependent, project-based funding that produces the "pilotitis" pattern of many small initiatives that never consolidate. For these systems, mobile-first design, offline-tolerant tools, and pooled regional infrastructure are more relevant than frontier AI.

A comparative reading across archetypes carries a clear lesson: wealth buys capacity but not coherence. The systems moving fastest on population-level digital services are often not the richest but those that treated digital infrastructure as a governed public good and settled the questions of payment and trust early.

“Wealth buys capacity but not coherence.”

5. Competitive landscape

The supply side mirrors the segment structure. In clinical IT, a small number of entrenched vendors — with Epic and Oracle Health prominent in the United States and a set of national and regional players elsewhere — hold durable positions built on high switching costs and deep clinical workflow integration. This concentration is a double-edged feature: it provides stability and standards gravity, but it also shapes the terms of interoperability and can slow the entry of complementary tools.

In virtual care, the post-peak period has been a shakeout. Several pure-play telehealth companies that scaled rapidly during the demand shock have faced margin compression and consolidation as usage normalised and reimbursement tightened; in parallel, virtual capability has been absorbed into incumbent provider systems and EHR platforms, blurring the line between "telehealth company" and "care delivered digitally." In Europe, booking-and-consultation platforms such as Doctolib illustrate a different, infrastructure-adjacent model.

Large technology firms occupy an ambiguous position. Apple, Alphabet, Amazon and Microsoft have each pursued health strategies — consumer sensing, cloud and analytics infrastructure, care delivery and ambient AI — with mixed results. The pattern to date is that platform and infrastructure plays (cloud, data, AI tooling sold to health organisations) have proven more durable than direct-to-consumer care ventures, several of which have been scaled back. Medical-device and imaging incumbents — Philips, GE HealthCare, Medtronic and peers — are extending into connected monitoring and embedded analytics, where their regulatory experience and installed base are advantages.

The AI layer is the most fluid. Regulators have authorised a growing catalogue of AI- and machine-learning-enabled medical devices — now numbering in the many hundreds and heavily weighted toward imaging and diagnostics — while a parallel, faster-moving wave of documentation and administrative AI diffuses largely outside device regulation. The competitive question over the next few years is whether value accrues to specialist clinical-AI firms, to the EHR incumbents that control the workflow and the data, or to the providers of the cloud and models underneath. The honest answer is that it is unsettled, and that control of clinical workflow and trusted data may matter more than model performance alone.

6. Risks that cut across the scenarios

Five risks could materially alter the trajectory, and we hold them separately from the scenarios because they cut across all of them.

Reimbursement reversal. The largest single uncertainty is policy, not technology. A withdrawal of permanent payment pathways for virtual and remote care — plausible under fiscal pressure — would return significant volume to a narrow set of use cases and strand investment predicated on continued growth.

Cybersecurity and resilience. Repeated ransomware incidents against hospitals and, notably, against health-data intermediaries whose failure can halt claims and prescriptions across a whole country, have established that digital dependence without resilience is a patient-safety exposure. As systems digitise further, the blast radius of a single failure grows. Resilience investment lags dependence almost everywhere.

Evidence and efficacy gaps. Much of the sector's forward valuation rests on outcomes that are asserted more often than demonstrated. Digital therapeutics in particular have seen reimbursement and evidence pathways prove fragile, with high-profile commercial setbacks. Weak or selectively reported evidence is both a clinical risk and a market risk.

AI governance and liability. As AI moves toward decision support, unresolved questions of liability, post-market monitoring of models that drift, bias in training data, and validation across populations become binding. Regulatory frameworks are maturing — the EU's risk-based approach to AI and evolving device-regulator guidance among them — but the gap between deployment pace and oversight capacity is a genuine risk.

Equity and the digital divide. Without deliberate design for the least-connected and least digitally literate, digital care will preferentially benefit those already best served. This is not only an ethical concern; it is a design and political-durability concern, because services perceived to widen inequity lose public support.

Honesty also means naming what we do not know. The durable margin structure of AI and analytics at scale is unproven; the long-run outcomes of substituting virtual for in-person care across the full range of conditions are not fully established; and the willingness of publics to sustain broad health-data sharing under commercial pressure is genuinely uncertain. Our figures reflect these unknowns as ranges, not point estimates.

Outlook to 2030

We offer three scenarios. They are not forecasts with assigned probabilities; they are internally consistent stories that bound the plausible range and clarify which decisions move a system between paths.

Embedded Care

Upside — durable integration

Payment pathways for virtual and remote care are made permanent and predictable; interoperability matures from format to meaning; AI moves under strengthened oversight from documentation into validated decision support; Leapfroggers consolidate national infrastructure.

Blended segment growth

Upper half of ranges

Binding requirement

Policy follow-through

Fragmented Acceleration

Central case — most likely

Adoption advances rapidly in pockets while reimbursement stays inconsistent, interoperability partial and equity gaps persist. Investment continues but is choppy, with periodic shakeouts in over-funded segments.

Blended segment growth

Middle of ranges

Distribution

Very uneven

Stalled Digitisation

Downside — setback

Reimbursement retrenchment, one or more severe cybersecurity or data-trust failures, and disappointing outcomes evidence erode momentum; capital retreats from the least-proven segments and digital delivery remains an adjunct.

Blended segment growth

Lower bound of ranges

Root cause

Governance & trust failure

Scenario A — Embedded Care (durable integration). Payment pathways for virtual and remote care are made permanent and predictable across most high-income systems. Interoperability mandates converging on FHIR mature from format to meaning. AI diffuses first through documentation and administration, then, under strengthened post-market oversight, into validated decision support. Leapfrogger systems consolidate national infrastructure and reach broad population coverage. In this path, blended segment growth runs toward the upper half of our indicative ranges, and digital delivery becomes an ordinary feature of care rather than a distinct category. The binding requirement is policy follow-through, not invention.

Scenario B — Fragmented Acceleration (uneven progress). The most likely path on current evidence. Adoption advances rapidly in pockets — specific specialties, specific well-funded systems, specific administrative use cases — while reimbursement remains inconsistent, interoperability stays partial, and equity gaps persist. Investment continues but is choppy, with periodic shakeouts in over-funded segments. Growth lands in the middle of our ranges, distributed very unevenly across and within countries. The sector matures in capability while under-delivering on the system-level coherence that would compound its value.

Scenario C — Stalled Digitisation (setback). A combination of reimbursement retrenchment under fiscal pressure, one or more severe cybersecurity or data-trust failures, and disappointing outcomes evidence erodes momentum. Public confidence in data sharing weakens; regulators tighten faster than the sector can adapt; capital retreats from the least-proven segments. Growth falls toward the lower bound of our ranges, clinical IT continues its slow expansion largely detached from the promised care-model change, and digital delivery remains an adjunct rather than a mainstay. This scenario is not a technology failure; it is a governance and trust failure.

Across all three, the mature clinical-IT base grows slowly and steadily; the difference between scenarios is concentrated in whether the faster-growing, less-proven segments convert their promise into durable, financed, trusted practice. The determining variables are legible and largely within policymakers' control.

Implications for each audience

Governments and policymakers

Treat digital infrastructure as a governed public good, not a procurement line item. The systems making the most population-level progress settled three things early: a stable payment mechanism, an interoperability mandate with teeth (converging on FHIR and shared terminologies), and a clear, trusted data-governance and consent regime. Prioritise these over frontier applications. Invest in cybersecurity resilience commensurate with the dependence you are creating, and design explicitly for the least-connected populations, because equity failures are also durability failures. Where you are a Leapfrogger, protect public trust as the scarce resource it is.

Business and investors

Underwrite to segments, not to the aggregate. The durable value in clinical IT lies in incumbency and interoperability; in virtual care, in reimbursement-anchored, clinically integrated models rather than standalone volume; in AI, in control of workflow and trusted data more than in model performance alone. Discount forward figures for the evidence and reimbursement risk that has already caught out parts of the digital-therapeutics and pure-play telehealth space. The most reliable near-term returns sit in tools that demonstrably remove clinician work and in the infrastructure layer beneath applications.

Providers and health systems

Adopt where the tool removes work and the payment pathway exists; resist adoption that adds documentation burden without offsetting benefit. Insist on interoperability and data portability in procurement to limit lock-in. Build workforce capacity deliberately — the constraint is rarely the software and usually the change management around it.

International organisations and donors

Fund consolidation over proliferation. The "pilotitis" pattern is a financing artefact as much as a technical one; pooled, multi-year support for shared regional infrastructure, standards adoption and workforce capacity will do more than another generation of isolated pilots. Support open standards and public-goods infrastructure that middle- and low-income systems can build on without vendor capture.

Method, data and status of the numbers

This report is a desk-based synthesis and analytical assessment, not a primary-data study. Our approach combines three inputs. First, established public data from multilateral and government sources — health-expenditure and coverage data from the World Health Organization and comparable bodies, connectivity statistics from international telecommunications data, national health-IT adoption reporting, and medical-device and AI authorisation records from major regulators. Where we cite these directions of travel — the scale of global health spending, the breadth of EHR adoption in high-income systems, the trajectory of connectivity, the growing catalogue of authorised AI-enabled devices — we rely on widely reported figures and describe them qualitatively rather than to spurious precision.

Second, a bottom-up market model for the segment sizing in Section 2. This reconciles public company disclosures, national IT-expenditure data and analyst syntheses into six segments, with deliberate adjustments to reduce double-counting across overlapping categories (for example, monitoring hardware embedded in telehealth programmes, or consumer devices with clinical-grade subsets). Third, comparative case reading of a purposive sample of systems spanning the four readiness archetypes.

We are explicit about the status of every number. The segment revenue ranges and all growth rates are planning estimates, expressed as ranges precisely because published totals diverge by a factor of two or more on scope alone; they should not be read as measured accounts. The forward figures and scenarios are conditional ranges, not point forecasts, and carry no assigned probabilities. Directional macro facts drawn from multilateral sources are the firmest ground in the report; the segment model is a transparent estimate; the scenarios are structured judgement. Readers should weight each accordingly. Limitations include the unevenness of cross-country data, the definitional instability of "digital health" itself, and the rapid movement of the AI segment, where any snapshot dates quickly.

This is independent, non-partisan analysis. The authors declare no commercial interest in any named vendor, and no funder had editorial control over the findings.

References

The multilateral, regulatory and standards sources below were read directly; company disclosures and analyst syntheses feed the bottom-up segment model and are referenced by category.

• World Health Organization (2021). Global Strategy on Digital Health 2020–2025. WHO, Geneva.
• World Health Organization (2022). Health workforce: global projections and the 2030 shortfall. WHO, Geneva.
• International Telecommunication Union (2024). Measuring Digital Development: Facts and Figures 2024. ITU, Geneva.
• OECD (2023). The COVID-19 Pandemic and the Future of Telemedicine, with health-IT indicators. OECD Publishing, Paris.
• Health Level Seven International (2023). HL7 FHIR Release 5 (Fast Healthcare Interoperability Resources). HL7 International, Ann Arbor, MI.
• US Food and Drug Administration (2024). Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices — authorised device list. FDA, Silver Spring, MD.
• European Parliament and Council of the European Union (2024). Regulation (EU) 2024/1689 (Artificial Intelligence Act). Official Journal of the European Union.
• National Health Authority, India (2023). Ayushman Bharat Digital Mission: architecture and progress report. Government of India, New Delhi.
• Company disclosures and analyst syntheses (2022–2025), used to build the bottom-up segment model and referenced by category.