The development of artificial intelligence in medical technology is usually discussed as a technical challenge – better algorithms, larger datasets, more computing power and closer integration with clinical practice.
For companies attempting to bring AI-enabled medical devices to market, however, another constraint is becoming increasingly important.
They need people who can translate technical innovation into clinical evidence, quality systems and a regulatory strategy that will withstand scrutiny throughout the product’s lifecycle and that combination of expertise is difficult to find.
In executive search, we have traditionally been asked to identify either technical specialists or experienced business leaders.
AI-enabled medical technology is creating demand for something more unusual: professionals who understand medical-device regulation but can also engage credibly with machine learning, software development, data quality, algorithmic performance and post-market monitoring.
The shortage is not simply a lack of regulatory-affairs professionals but rather global lack of people who can work effectively across disciplines that have historically developed separately.
A narrow shortage within a wider skills problem
In a recent report Life Sciences 2035: Developing the Skills for Future Growth, a group of UK life-sciences organisations – including the Association of British HealthTech Industries – estimated that the sector could require 70,000 additional jobs by 2035, as well as 75,000 people to replace those leaving the workforce.
Those figures cover the broader life-sciences industry rather than AI regulation specifically, but they provide important context. Companies are trying to recruit highly specialised AI regulatory professionals from a labour market that is already under pressure.
The required expertise is so unusually broad that few, if any, people have developed all of the necessary capabilities in a single career.
They include understanding how the intended purpose of a software determines whether it is regulated as a medical device; the quality and representativeness of training and validation data; and the clinical evaluation and the evidence needed to support performance claims.
They also need to understand the nature of bias, explainability, human factors and the limits of model outputs; software lifecycle processes, cybersecurity and change control; performance drift and post-market monitoring; and the different regulatory expectations of the markets in which the product will be sold.
Experienced regulatory professionals may have spent decades working with conventional devices, diagnostics or pharmaceuticals without acquiring a detailed understanding of machine-learning development.
Conversely, software and data-science specialists may have little experience of clinical evidence, quality-management systems or regulated product development. The valuable people are those who can translate between the two groups.
Regulation must be designed into the product
When regulatory expertise is introduced late, companies may discover that the data collected during development are insufficient, that performance claims cannot be supported, or that changes to an algorithm cannot be managed within the proposed quality system.
In more serious cases, the product may require additional studies, substantial redesign or a revised route to market. This can delay commercialisation and consume capital that was originally intended for product development, market expansion or clinical adoption.
Regulatory strategy should therefore be treated as part of product design rather than as a submission exercise conducted after development has been completed. The regulatory professional must be involved early enough to influence the intended purpose, development plan, evidence strategy, risk-management process and arrangements for monitoring the product after deployment.
A more complex international environment
The challenge is intensified by differences between regulatory systems. In the European Union, for example, AI-based software intended for medical purposes can fall within the high-risk provisions of the AI Act while also being subject to the Medical Devices Regulation or the In Vitro Diagnostic Medical Devices Regulation.
The European Commission identifies risk management, data quality, information for users and human oversight among the requirements that apply to high-risk healthcare AI.
The AI Act also requires providers and deployers of AI systems to take measures to ensure an appropriate level of AI literacy among relevant staff. That obligation has applied since 2 February 2025, ahead of the later phases of the Act’s implementation.
Compliance therefore involves more than appointing a single expert. Organisations must develop a wider understanding of AI among the people responsible for designing, evaluating, deploying and overseeing these systems.
The UK is following a different route. Rather than introducing one comprehensive cross-sector AI law, it has generally asked existing regulators to apply common principles within their own areas of responsibility.
For medical technology, the Medicines and Healthcare products Regulatory Agency is developing its Software and AI as a Medical Device Change Programme to clarify requirements and protect patients.
The MHRA has also used its AI Airlock regulatory sandbox to examine some of the distinctive difficulties associated with AI medical devices. Its pilot work underlines the need for regulators, developers and healthcare professionals to learn together as the technology evolves.
The competitive question is therefore not which market has the least regulation. It is which companies can understand and navigate the applicable requirements most effectively.
The commercial effect of scarcity
Demand for professionals who combine regulatory experience with meaningful AI, software or data expertise now exceeds the immediately available supply.
Scarcity affects both salaries and access to talent. Large medical-technology companies and established consultancies are often better placed to offer higher compensation, specialist colleagues and a succession of technically demanding projects. Smaller businesses may struggle to compete, despite sometimes having the most urgent need for the expertise.
This has helped create a strong market for specialist consultants. External support can be valuable, particularly when a business needs expertise in a particular jurisdiction, submission or stage of development. But consultancy does not remove the need for internal ownership.
Building capability rather than searching for unicorns
Companies need to build multidisciplinary regulatory capability. In practice, that may mean combining an experienced medical-device regulatory leader with specialists in software, data science, clinical evaluation, cybersecurity and quality assurance rather than expecting one person to possess expert knowledge in every field.
Existing regulatory professionals also need structured opportunities to develop their understanding of AI. This does not require every regulatory leader to become a data scientist. It does require them to understand how models are developed and validated, where bias or performance drift can arise, how changes affect the approved product and what evidence is needed to demonstrate continuing safety and effectiveness.
Longer-term training initiatives will help broaden the pipeline. New Higher Technical Qualifications being developed through partnerships involving Skills England, employers and education providers are intended to support roles associated with AI-enabled health devices and regulatory advice. They will not resolve the immediate shortage of senior professionals, but they represent the type of practical, cross-disciplinary training that the sector will increasingly require.
Regulatory professionals must also make intelligent use of AI themselves. Appropriate tools can assist with document analysis, evidence management, regulatory intelligence and routine monitoring. Used responsibly, they can release experienced people to concentrate on judgement, strategy and communication.
The decisive issue is not whether regulation will slow the development of AI in MedTech. It is whether companies can build the regulatory capability required to develop, validate and commercialise these products responsibly.
The technology may be advancing rapidly. The organisations most likely to benefit from it will be those that develop the human expertise to keep pace.
Share