AI systems are only as good as the procurements that select them

Just over two years ago, I experienced what a healthcare procurement process can mean in practice. At the time, I was working for a company in the field of orthotics and prosthetics, and when the procurement contract was lost, the conditions for our business changed from one day to the next. The experience led me to a realization: procurement is not only about law, processes, and economics. It also determines which products, which services, and ultimately which care patients gain access to.

Public procurement serves an important function. It should be legally secure, transparent, and give all suppliers the opportunity to compete on equal terms. However, if too much emphasis is placed on price alone, there is a risk that other values will be overshadowed. In healthcare, this can have consequences for the quality of the tools used in patient care. Procurement needs to balance different values against one another, including values that are not always easy to measure.

This was one of the starting points for a recent article that my research colleague Santa Slokenberga and I published. In the article, we examine how European legislation can contribute to more balanced healthcare procurements, procurements that do not risk overemphasising certain values, such as price, at the expense of other important considerations, such as quality and safety. The article focuses on the procurement of AI systems, but many of our conclusions are more broadly applicable.

We explore the interaction between three key European legal frameworks: the Artificial Intelligence Act (AIA), the Medical Device Regulation (MDR), and the Public Procurement Directive (PPD). Using Donabedian’s model of healthcare quality, we analyse how these legal frameworks contribute to quality in terms of healthcare structures, processes, and outcomes.

Our conclusion regarding AI systems is that the AIA and MDR contain important requirements related to areas such as safety, transparency, documentation, risk management, and post-market monitoring. However, legislation alone is not enough to ensure that AI actually contributes to better care for patients. These requirements only acquire real significance when they are translated into practice, and procurement can play an important role in this respect. We argue that procurement processes which take the requirements of the AIA and MDR into account can create favourable conditions for high-quality healthcare. Through procurement, healthcare providers can set requirements that go beyond CE marking and formal regulatory compliance. This may include requesting evidence of clinical benefit, requirements concerning diagnostic accuracy, follow-up of patient outcomes, transparency about how systems have been developed, and how they perform in the environment in which they are intended to be used. In other words, procurement that takes these two regulations into account can serve as a tool for translating legal requirements into actual quality of care.

An important message of the article is that procurement processes cannot be isolated from clinical reality. Procurement professionals need support from clinicians, patients, technical experts, and other relevant stakeholders to formulate meaningful quality requirements. If healthcare is to benefit from AI, those who meet patients must be involved in defining what quality actually means. AI should be implemented not simply because there are political, clinical, or public expectations that everything will improve with AI, but only when the procured AI system genuinely addresses clearly identified needs in accordance with relevant requirements.

Technology is developing rapidly and offers significant opportunities. But technology should not set the agenda. It is we, as humans, who must define the framework, establish the requirements, and determine which solutions deserve a place in healthcare. Only then can AI become a tool that strengthens, rather than steers, the future of healthcare.

If you would like to read more, you can find our article here: A Pathway from Law to Care: How the Artificial Intelligence Act, the Medical Device Regulation, and the Public Procurement Directive Can Contribute to Ensuring Quality in Healthcare.

Many of our conclusions regarding the procurement of AI systems can be generalised. Procurement that takes seriously the task of balancing different values against one another, in accordance with relevant legal requirements, can provide a concrete pathway to both cost-effective and high-quality healthcare.

Viberg Johansson, J., Slokenberga, S. A Pathway from Law to Care: How the Artificial Intelligence Act, the Medical Device Regulation, and the Public Procurement Directive Can Contribute to Ensuring Quality in Healthcare. Health Care Analysis (2026). https://doi.org/10.1007/s10728-026-00588-9

This post in Swedish

Approaching future issues