Health economic insights from Hardian Summit 2026
Software and AI medical devices (AI/SaMD) are solving clinical problems at speed. Healthcare systems, though, won't adopt a technology that works brilliantly if they can't make an economic case for it. At the Hardian Summit 2026, we hosted a workshop on Health Economics and Outcomes Research for professionals new to the field. The attendees came in believing overwhelmingly that investors in AI/SaMDs need economic evidence or conceptualisation of impact beyond the device working. Ninety-six percent said yes. More than three-quarters called it a necessity. HEOR is how that case gets built.
We are keen to share the key insights and discussions from this session.
What is HEOR and what are its benefits?
We described HEOR as a structured framework that enables decision-makers to make informed decisions regarding the reimbursement of health technologies. By evaluating all available evidence, this approach estimates the clinical and economic impact of health technologies on patients, healthcare providers, and society as a whole. The key aspects of HEOR are that:
HEOR evidence generation happens via health economic evaluations, a key approach for identifying technologies with the greatest clinical and economic value for stakeholders and patients to whom AI/SaMDs would matter the most under increasingly constrained healthcare budgets.
Budget impact models, cost-effectiveness models and multi-criteria decision analyses are the main approaches for economic evaluations of AI/SaMDs, though they vary in terms of audience, time horizon, outcomes required for collection and modelling complexity. While cost-effectiveness models target HTA agencies over the long term, budget impact models use shorter time horizons for potential buyers, such as healthcare providers. Meanwhile, multi-criteria decision analyses focus on the economic perspective of a chosen clinical stakeholder at a static point in time, while they also allow greater flexibility for inclusion of outcomes.
HEOR data collection can be highly beneficial for different stakeholders, including traditional HTA agencies, customers and investors of AI/SaMDs, as well as manufacturers of AI/SaMDs, for different reasons:
HTA agencies: HEOR can help demonstrate the potential clinical and economic value of a new technology within a healthcare system, comparing it to current clinical practice and other available technologies. HEOR also helps with addressing evidence gaps , as it identifies and resolves missing evidence, data points, or barriers that could hinder commercialisation and wide-scale adoption.
Customers and investors of AI/SaMDs: HEOR data collection allows for informed decision making based on the best available evidence, identification of cost savings when budgets are constrained, and lower risk for investment.
Manufacturers/developers of AI/SaMDs: HEOR is capable of identifying an appropriate benchmark price for such technologies, choosing the optimal use case that can generate the highest value to potential buyers, and helping manufacturers with accessing research grants that could help support with development and commercialisation as health economics often forms part of the grant application criteria.
During the session, we asked workshop attendees on whether they believe that investors need evidence or conceptualisation of impact beyond the device working. The feedback was overwhelmingly positive (96%), with 74% of respondents identifying this as a necessity and 22% noting it would be advantageous to have.
Relevant metrics for AI/SaMD economic evaluations
Having outlined the benefits of HEOR for AI/SaMDs, the workshop continued by outlining the main clinical, patient-reported, and health economic outcomes used in economic evaluations of AI/SaMDs in radiology. These metrics, also relevant for any diagnostic devices or intervention pre-diagnosis, can serve as benchmarks for developers aiming to identify a persuasive health economic case for their own AI/SaMD technologies before wider-scale commercialisation and market adoption.
Key clinical metrics may include:
Length of stay
Treated or not treated
Time to treatment
Time to diagnosis
Frequency of investigation
Complications or adverse events
Image quality
Diagnostic accuracy
Relevant patient-reported outcomes may involve:
Health-related quality of life
Functional independence
Health economic outcome metrics consist of three main aspects:
Direct costs, including administrative costs, screening costs, diagnostic costs, therapeutic costs, litigation costs and software intervention costs.
Productivity costs, including clinician’s reporting time, staff time spent on completing tasks, inter-reader variability or reliability and scan reading time.
Cost-effectiveness outcomes, such as incremental cost per QALY gained or DALY averted, as well as incremental cost per deaths, hospitalisations, or inaccurate diagnoses diverted.
These findings stem from a peer-reviewed systematic review published in European Radiology, led by Hardian Health in partnership with Prof. Zanca of the European Commission. In summary, economic evaluations in radiology AI are increasingly prevalent, reporting diverse economic data with a primary focus on productivity, cost-effectiveness and cost reduction. Explore our Hardian blog or access the full peer-reviewed article for more details on our systematic review.
What are the challenges with HEOR data collection?
While the advantages of HEOR data collection are multiple and substantial, our extensive engagement with diverse stakeholders has highlighted several challenges related to HEOR data collection. Addressing these barriers effectively requires close coordination among all parties early to streamline the evidence generation process and ensure a robust health economic evaluation of an AI/SaMD.
Downstream data collection is the most frequent challenge encountered in HEOR projects. To address this challenge, alternative data collection approaches can be implemented including comprehensive literature reviews, interviews with key clinical stakeholders, and prioritising data collection on first-order impact outcomes. Clinical stakeholders should be involved at the start of the project to reach a consensus on health economic outcomes, as there might be variation in choosing which of such outcomes to prioritise for data collection.
While data are collected from clinical sites, there may be variations in data metrics. To streamline the data collection process, a clear list of inputs can be designed before data collection, while a harmonisation strategy can be developed with contributions from both health economists and AI/SaMD manufacturers. Another source of variation in clinical sites could be related to the heterogeneous clinical management for the disease explored in the health economic evaluation. To account for variations in clinical practice, health economists should design and conduct health economic models capable of handling different scenarios.
The HEOR data collection process can be time consuming and costly, an aspect that can jeopardise the progress of the project given existing time and financial budget constraints. All parties should reach consensus on a comprehensive data gathering strategy, as well as identify risks and their mitigations at the early stages of a project.
If patient-level data is collected for HEOR evidence generation, ethical obligations may be involved in the processing of such data. To help address this challenge, health economists can work alongside clinical stakeholders and manufacturers to receive a list of relevant descriptive statistics if anonymised individualised data are unavailable.
Finally, manufacturers may have an unclear understanding of the economic potential of their AI/SaMD technology, especially at the early stages of their product development. Early economic modelling is a valuable tool to help identify the drivers of economic value for AI/SaMDs before wide-scale adoption and evidence generation.
During the session, we asked workshop attendees on the HEOR data collection challenges they have encountered so far. Whereas downstream data collection was reported to be the most prevalent challenge (64%), most other challenges were also significant with at least 35% of survey respondents experiencing them, while data sharing with third-party analysts and evaluators was reported less frequently at 21%.
Concluding remarks
Considering the multidimensional benefits, aspects and challenges associated with HEOR data collection, the HEOR data collection process should be planned well in advance in order to lead to robust evidence generation and, subsequently, to a strategic commercialisation pathway. Teams involved in economic evaluations of AI/SaMDs are encouraged to:
Identify relevant clinical stakeholders to establish a clinical stakeholder consortium, which could support a targeted HEOR data collection.
Ensure appropriate timeframe is allocated towards conceptualising the potential economic value of the AI/SaMD before health economic modelling begins.
Produce a robust data gathering strategy, accounting for potential risks. Such a strategy should list assumptions, inputs and different scenarios reflecting real-world clinical practice.
Identify the direct cost of the intervention, which may consist of training, implementation, subscription and maintenance costs.
Estimate accurately the clinical performance of the AI/SaMD based on its specified intended use.
Be aware of the data infrastructure of clinical sites and its implications for data collection.
Critically appraise and evaluate economic models to ensure the findings are clear, transparent and robust.
Account for uncertainty of the economic model’s results, while also performing scenario analyses reflecting heterogeneous clinical practice if applicable.
We hope these considerations and tips regarding HEOR data collection are helpful for your journey towards evidence generation and market adoption. Provided that proactive risk mitigations and data collection strategies are established at the early stages of a project, HEOR can become a powerful tool for determining the economic value of AI/SaMDs and , consequently, an informant of commercial strategy.
