High-Risk Software (III–IV) as Medical Device Market By Product Type (Diagnostic Interpretation Software, Clinical Decision Support Software, Patient Triage and Prioritisation Software, Monitoring and Risk-Prediction Software, Treatment-Planning Software, Autonomous Diagnostic Software); By Application (Radiology and Diagnostic Imaging, Cardiology, Neurology and Stroke Care, Oncology, Ophthalmology, Pathology, Emergency and Critical Care); By End User (Hospitals and Health Systems, Diagnostic Imaging Centres, Specialty Clinics, Ambulatory and Outpatient Care Centres, Public Healthcare Organisations, Medical-Device and Healthcare-Software Companies); By Geography, Segment Revenue Estimation, Forecast, 2026–2032

High-Risk Software (III–IV) as Medical Device Market : Clinical Intelligence Moves from Regulatory Approval to Recurring Revenue

The Global High-Risk Software (III–IV) as Medical Device Market was valued at USD 2.55 billion in 2025 and is projected to reach USD 6.50 billion by 2032, expanding at a CAGR of 14.3% during 2026–2032, according to Strategic Market Research.

 

High-risk SaMD platforms are increasingly integrated with drug development and clinical treatment pathways by using AI-driven insights to support accurate diagnosis, patient stratification, and therapy selection. These systems help clinicians identify suitable drug regimens, predict treatment responses, and optimize interventions in critical areas such as cardiology, neurology, and oncology. AI-based diagnostic tools can also monitor disease progression and guide personalized medicine approaches, improving the safety and effectiveness of therapies. Emerging solutions are enabling faster clinical decisions and enhancing the use of targeted drugs in complex and life-threatening conditions.

 

When Clinical Decisions Carry Greater Risk, Regulatory and Commercial Barriers Rise

The IMDRF framework classifies SaMD according to the seriousness of the health condition and the importance of the information supplied. Category IV includes software used to diagnose or treat critical conditions, while Category III includes software used to diagnose or treat serious conditions or drive the clinical management of critical conditions.

Higher-consequence products require stronger clinical evidence, quality controls, cybersecurity safeguards, post-market monitoring and governance than routine workflow applications. Suppliers must demonstrate reliable performance across intended populations and clinical environments, while health systems frequently require local evaluation before broad deployment.

 

Imaging-Led Workflows Build the Market’s Largest Recurring Revenue Engine

Radiology and diagnostic imaging provide the strongest concentration of high-consequence SaMD authorisations, procurement programmes and recurring clinical transactions. In January 2025, the FDA reported that more than 1,000 AI-enabled medical devices had been authorised through established premarket pathways. Its public list includes decisions through 30 March 2026, with substantial activity in radiology, cardiovascular care, neurology and treatment planning.

NHS UK reported 49.9 million imaging examinations in the year to March 2025, representing 5.9% growth from 47.2 million in 2023/24. X-rays accounted for 23.5 million examinations, or 47.1% of the total. Ultrasound represented 11.5 million, or 23.0%; CT contributed 8.3 million, or 16.6%; and MRI contributed 4.9 million, or 9.8%. These volumes create recurring demand for interpretation, prioritisation, quantitative analysis and clinical decision support.

Every eligible scan or patient assessment can generate revenue under a per-analysis arrangement, while subscription contracts convert the same activity into annual recurring revenue. Suppliers integrated into radiology systems, electronic health records and reporting workflows are better positioned to improve utilisation and renewal rates.

 

Public Healthcare Funding Converts Clinical AI Pilots into System-Wide Deployment

UK offers clear evidence that public funding can move clinical AI from pilots into multi-site deployment. In October 2023, the UK government allocated £21 million to 64 NHS trusts for AI tools supporting chest X-ray and CT analysis in lung-cancer pathways. The programme linked adoption to a defined clinical priority, large examination volumes and pressure to shorten diagnostic waiting times.

Stroke decision support provides an even stronger adoption signal. NHS England reported that AI decision-support use increased from 5% of stroke units in 2019 to 90% by August 2023, an 85-percentage-point rise and eighteenfold expansion. Acute-stroke software can prioritise suspected large-vessel occlusion cases and alert specialists, supporting faster decisions on thrombolysis, thrombectomy or specialist transfer.

The NHS AI in Health and Care Award allocated more than £100 million between 2020 and 2024. Thirteen Phase 4 technologies entered multi-site deployment and independent real-world evaluation. Scaled commissioning therefore depends on clinical evidence, operational value, implementation capacity and performance in routine care.

 

Health-System Consolidation Reshapes Enterprise Sales and Contract Economics

The United States had 6,100 hospitals based on FY2024 data, including 5,121 community hospitals, which represented 84.0% of the national hospital base. Of these, 3,567 were affiliated with healthcare systems, equivalent to 69.7% of community hospitals and 58.5% of all US hospitals. The AHRQ Compendium separately identified 639 US health systems containing at least one hospital and one physician group.

A health-system contract may cover several hospitals, outpatient centres, imaging locations and specialist practices. This structure allows suppliers to scale through fewer strategic accounts, but it also gives buyers greater negotiating power. Enterprise customers expect interoperability, cybersecurity documentation, implementation support, measurable performance and network-wide pricing.

Demand usually begins with delayed image interpretation, specialist shortages, increasing diagnostic volume or a time-critical treatment pathway. Revenue conversion requires alignment among clinicians, IT teams, security functions, procurement, finance, clinical governance and, where a separate service is billed, the payer. Sales cycles are long, but successful integration can support renewals and raise switching costs.

 

Reimbursement Unlocks Premium Per-Analysis Revenue in High-Value Clinical Pathways

Heartflow provides the clearest public example of a scaled high-consequence SaMD business linked to clinical use and per-analysis revenue. Its platform analyses coronary CT angiography data to quantify coronary anatomy, plaque and the effect of disease on blood flow, supporting decisions on further investigation, medical management and possible revascularisation.

Heartflow generated US$176.0 million in revenue during 2025, an increase of 40% year over year and equivalent to approximately 6.9% of the global market value. US revenue reached US$160.6 million, accounting for 91.3% of company revenue and approximately 6.3% of the global market. Its platform had been deployed in more than 1,465 US accounts by the end of 2025.

FFRCT represented 98% of Heartflow’s 2025 revenue, equal to approximately US$172.5 million and about 6.8% of the global market value. In the first quarter of 2026, revenue increased 41% to US$52.6 million and gross margin reached 80.2%. The company also reported that its technology had supported care for more than 650,000 patients worldwide.

Reimbursement remains application-specific. CMS quality specifications recognise autonomous diabetic-eye examinations billed under CPT 92229, while code 75580 supports administrative pathways for non-invasive coronary fractional-flow-reserve analysis in certain settings. Dedicated coding strengthens revenue conversion by linking software use to a billable clinical service.

 

Platform Consolidation Shifts Competitive Power Beyond Stand-Alone Algorithms

A second major revenue model is developing around enterprise platforms that combine multiple clinical algorithms, workflow tools, cloud delivery, support and data orchestration. RadNet’s Digital Health segment generated US$92.7 million in 2025 revenue, up 41.1%, equivalent to approximately 3.6% of the global 2025 market. Annual recurring revenue reached US$75.4 million, representing 81.3% of segment revenue and approximately 3.0% of the global market value.

The segment’s customer base increased from 486 to 2,075 during 2025, representing growth of approximately 327%. Reported procedure volume rose from 19.1 million to 24.1 million, an increase of 26.2%, with much of the expansion linked to the iCAD acquisition. Following the Gleamer transaction announced in March 2026, DeepHealth and Gleamer reported a combined installed base exceeding 2,700 customer contracts across more than 50 countries.

 

Approved SaMD Platforms Reveal Multiple Routes from Clinical Utility to Revenue

Heartflow represents a premium cardiovascular model in which repeated analysis volume and reimbursement support usage-based revenue. Its US$176.0 million 2025 revenue, 40% annual growth and estimated 6.9% equivalent share of the total market provide the clearest public benchmark for a high-value coronary CTA analysis platform.

Viz LVO, previously authorised through the De Novo pathway as ContaCT, analyses head CT angiography for findings suggestive of large-vessel occlusion and sends notifications to relevant specialists. Its value lies in reducing communication and transfer delays across stroke networks. England’s increase from 5% to 90% stroke-unit AI adoption demonstrates the scale that time-critical triage software can achieve.

iCAD’s ProFound AI analyses digital breast-tomosynthesis images for suspicious soft-tissue densities and calcifications. It supplies lesion and case scores that support radiologist interpretation and clinical management. Its inclusion within RadNet’s expanding platform reflects movement from stand-alone breast-imaging algorithms toward broader contracted imaging-AI portfolios.

 

Lifecycle Regulation Becomes a Strategic Moat for Well-Resourced SaMD Suppliers

High-risk SaMD suppliers compete on the quality of their full product lifecycle. Performance monitoring, change control, cybersecurity, data quality, bias assessment and post-market evidence increasingly affect procurement and contract renewal.

FDA’s August 2025 final guidance on predetermined change-control plans allows manufacturers to propose future AI-enabled software modifications and the methods for developing, validating and implementing them. An accepted plan can reduce regulatory friction for listed changes while preserving validation and monitoring obligations.

IMDRF finalised N81 in January 2025 to address medical-device software characterisation and software-specific risks. In the European Union, Rule 11 places software in Class III when decisions may lead to death or irreversible deterioration and in Class IIb when decisions may cause serious deterioration or require surgical intervention. Relevant products may also be subject to MDR or IVDR requirements and the EU AI Act.

The United Kingdom is expanding regulatory capacity through the MHRA AI Airlock programme, supported by £3.6 million over three years from 2026. Its work addresses explainability, changing model performance, synthetic data, hallucination risk and user over-reliance.

 

Regional Growth Models Diverge Across Reimbursement, Public Procurement and Regulation

The United States offers the clearest evidence of usage-based monetisation, dedicated coding and specialist enterprise adoption. Heartflow generated US$160.6 million of its 2025 revenue in the US, equivalent to 91.3% of company revenue. This concentration supports the country’s position as the most commercially visible market for reimbursed high-consequence analysis.

England shows how central funding and national priorities can support coordinated deployment. Lung-imaging AI received £21 million for deployment across 64 trusts, stroke decision support reached 90% of stroke units by August 2023, and national imaging activity grew 5.9% in 2024/25.

The European Union presents a large multi-country opportunity with substantial regulatory barriers. Conformity assessment, Rule 11 classification and the combined application of medical-device and AI regulation increase market-entry costs. Asia-Pacific is gaining importance as hospitals expand digital imaging, AI-supported diagnostics and specialist-care capacity, although progress will vary by regulation, hospital IT maturity, reimbursement and procurement support.

 

The Clearance-to-Utilisation Gap Remains the Market’s Largest Commercial Risk

The largest revenue leakage occurs between regulatory authorisation and paid utilisation. A product may receive clearance but fail to secure a hospital budget, tender award, reimbursement pathway or enterprise contract. Purchased software may then face delays during cybersecurity review, integration, local validation and clinical-governance approval.

Deployment does not guarantee routine ordering. Clinicians may use the tool inconsistently, distrust its output or encounter patients and studies outside its intended use. Only part of the eligible workflow may convert into paid analyses.

Heartflow’s 98% dependence on FFRCT, representing approximately US$172.5 million in 2025 revenue, provides focus and strong economics but increases exposure to changes in coverage, guidelines, competition and coronary-CTA utilisation. Portfolio strategies reduce dependence on one application but introduce acquisition and integration challenges. RadNet’s 327% customer-base expansion compared with 26.2% procedure-volume growth also shows that acquisition-led account growth does not translate proportionately into utilisation.

 

From Algorithm Sales to Embedded Clinical Revenue: The Market’s Strategic Direction

The High-Risk Software (III–IV ) as Medical Device Market is moving from stand-alone algorithm sales toward clinically embedded, recurring revenue models. Radiology remains the most visible opportunity because it combines large digital workflow volumes, established data formats and active regulatory pipelines. X-rays represented 47.1% of NHS England’s 2024/25 imaging activity, ultrasound 23.0%, CT 16.6% and MRI 9.8%, giving suppliers several procedure pools for prioritisation, interpretation and analysis.

Acute-stroke decision support demonstrates how time-critical value can drive adoption, while coronary CTA analysis shows the premium economics available under reimbursed per-case models. Heartflow’s US$176.0 million revenue represented approximately 6.9% of the 2025 global market, while RadNet Digital Health represented approximately 3.6%. RadNet’s US$75.4 million ARR accounted for 81.3% of its segment revenue, confirming the increasing importance of contracted recurring income alongside per-analysis models.

 

High-Risk Software (III–IV) as Medical Device Market Report Coverage Table

Report Attribute	Details
Forecast Period	2026 – 2032
Market Size Value in 2025	USD 2.55 Billion
Revenue Forecast in 2032	USD 6.50 Billion
Overall Growth Rate	CAGR of 14.3% (2026 – 2032)
Base Year for Estimation	2025
Historical Data	2019 – 2024
Unit	USD Million, CAGR (2026 – 2032)
Segmentation	By Product Type, By Application, By End User, By Region
By Product Type	Diagnostic Interpretation Software, Clinical Decision Support Software, Patient Triage and Prioritisation Software, Monitoring and Risk-Prediction Software, Treatment-Planning Software, Autonomous Diagnostic Software
By Application	Radiology and Diagnostic Imaging, Cardiology, Neurology and Stroke Care, Oncology, Ophthalmology, Pathology, Emergency and Critical Care
By End User	Hospitals and Health Systems, Diagnostic Imaging Centres, Specialty Clinics, Ambulatory and Outpatient Care Centres, Public Healthcare Organisations, Medical-Device and Healthcare-Software Companies
By Region	North America, Europe, Asia-Pacific, Latin America, Middle East and Africa
Country Scope	U.S., Canada, UK, Germany, France, Italy, China, Japan, South Korea, India, Brazil, Mexico, Saudi Arabia, UAE, South Africa
Market Drivers	• Rising adoption of AI-enabled clinical decision support and autonomous diagnostic technologies • Increasing regulatory acceptance of high-risk Software as a Medical Device solutions • Growing demand for workflow automation, precision diagnostics, and data-driven healthcare delivery
Customization Option	Available upon request