Bevan Brittan’s Vincent Buscemi, Commercial Healthcare Partner and Head of Independent Health and Care, Brett Lambe, Technology and IP Partner, and solicitors Aisha Matoo and Yasmin Cant, consider why regulation in AI is important
In recent years, we have seen incredible technological advancements due to the creation and roll-out of Artificial Intelligence (AI). From OpenAI’s ChatGPT answering complex questions in seconds (albeit with varying degrees of accuracy) to the NHS implementing AI technology to help diagnose COVID-19 from chest imaging, the range of use cases for AI in healthcare is vast and continues to grow as the technology develops.
Whilst the integration of AI in healthcare will undoubtedly transform the industry for providers and practitioners, and most importantly the lives of patients receiving care, the potential pitfalls, including ethical, regulatory and safety concerns, raise the risk of causing significant harm if not adequately addressed.
Lawmakers are working at pace to draft legislation and guidance to protect the systems and people using them; however, progress has been slow and trails far behind the pace of AI development, as is so often the case when the cultures of rapid technological innovation collide with the methodical caution of lawmakers. Even in Europe, where the EU’s AI Act was passed into law last year, widespread implementation has been slow; one year on from its approval by the EU Council, MedTech Europe was still urging policymakers to implement the legislation in cohesion with existing laws to encourage continued innovation. Beyond Europe, governments across the world are using a patchwork of historic legislation for existing technology to govern the development and use of AI in healthcare.
This discrepancy in the nature, scope and reach of AI laws at a macro level makes it difficult for technology providers to plan ahead, if they are looking to “bake in” compliant processes to their AI tools. Similarly, healthcare providers and practitioners are exercising caution in their deployment of AI, wanting to balance the benefits to patients with the need to remain compliant with current (and prospective) legal, clinical and ethical frameworks.
Data integrity and information security are integral to this challenge. Alongside patient safety and quality of care, protecting the data inputted into the AI models, for training and general use, is paramount. Though used across the world, most AI systems are developed, trained and tested in a small handful of countries. Consequently, large data groups from around the globe are often excluded from this training and testing, which can result in output bias and an increase in the health equality gap.
Lower-income countries do not have the same expertise and funding to contribute to the development of these systems, meaning higher-income countries pull further ahead. In addition to increased bias and wider health inequalities, vital patient data is regularly transferred across jurisdictions governed by differing data protection and privacy laws. This irregularity adds further to the complex matrix of distinct but interconnected issues that need to be balanced, with technology providers and major health governing bodies looking to scale at a global level, but needing to operate within separate and frequently inconsistently (or even incompatibly) regulated national healthcare markets.
Whilst we should continue to push for effective, pragmatic legislation at the national level, to reap the benefits of AI at the scale needed to be truly transformative we need to start looking beyond single-state national interests in isolation and consider solutions that can operate worldwide. The delivery of global benefits requires a global system, global co-operation and a global response.
One model for coordinating national systems to deliver a global approach may be provided by HealthAI.
HEALTHAI: A case study
HealthAI is an independent, non-profit organisation that aims to promote equitable access to and responsible governance of AI. It collaborates with governments, organisations and health leaders globally to support countries in building robust AI regulation.
With a focus on unlocking the transformational potential of AI, it is conscious of the disparity amongst countries adopting emerging technologies in the healthcare space. It addresses the need to ‘bridge the gap’ between low- and middle-income countries that lack the same levels of access to AI as their global counterparts. Peer-to-peer learning among countries should not only bridge the gap, but also hopefully increase the range and quality of overall data needed to inform optimum outcomes for global health challenges.
The pace of evolution of AI technology is clear. It is therefore critical to recognise the pressing need to govern and regulate the adoption of AI in healthcare at a global level. HealthAI asserts its commitment to facilitating a global function to regulate AI in healthcare, thereby mitigating risks and reducing the costs associated with integrating AI into healthcare systems.
Key to this approach will be implementation of proven global standards in practice, opening the door for AI-powered health technologies to be validated against these standards more quickly. If rolled out successfully, these criteria will enable countries to harmonise and expedite the certification of transformative technologies through standards-based validation.
Global regulatory network
A key component of HealthAI’s responsible AI governance and regulation is the establishment of the ‘Global Regulatory Network’. This network is designed to foster collaborative learning and sharing of best practices across different countries. It comprises ‘pioneering’ countries, enabling them to engage and collaborate with HealthAI to strive towards the common goal of improving and safeguarding the future of global health and wellbeing.
Earlier this year, the UK became the first ‘pioneer’ country to join the Global Regulatory Network, having been invited to participate by HealthAI. The Medicines and Healthcare products Regulatory Agency (MHRA) is the regulator of medicines and medical devices in the UK. With the UK being the first national member of HealthAI’s network, it positions the MHRA at the centre of the global efforts to regulate trusted and certified AI tools safely for frontline use. Singapore and India have also attained ‘pioneer country’ status by joining the network. IndiaAI and the Indian Council of Medical Research’s National Institute for Research in Digital Health and Data Science will collaborate with HealthAI to monitor the use of AI tools in practice, providing key clinical insights to inform further knowledge sharing among stakeholders across multiple countries.
Delving deeper into the Global Regulatory Network, HealthAI is targeting the development of a ‘global early warning system’. This system will be designed to detect and notify the network at the earliest stages of significant harm caused by AI software, such as malfunctions and software changes, allowing countries to identify and respond to harm promptly. While regulation is critical, it is arguably more important to establish such preventative systems, given the broad scope and application of AI in healthcare and the potentially severe impact of any unintended consequences. However, the development of technology will always outpace the implementation of legislation.
An additional tool in regulating the AI landscape by HealthAI is the creation and maintenance of a global directory of regulatory information on AI-related and registered medical devices (e.g., FDA-cleared or CE-marked devices) into a single, transparent and globally accessible database. This move should help countries in reviewing and evaluating registered AI healthcare tools and solutions for adoption in their localities.
As an example of the interconnected nature of national and international regulatory alliances, HealthAI’s CEO Dr Ricardo Baptista Leite sits on the UK’s newly launched National Commission into the Regulation of AI in Healthcare. This commission is aimed at supporting the MHRA in establishing a new regulatory framework. The framework is envisaged to be published in 2026 and will be the product of collaboration with global leaders and clinicians.
National standards are clearly essential, but using these national standards as a foundation for developing global standards is crucial for consistent and cohesive international AI regulation in healthcare. Acknowledging the existing issues associated with using AI in healthcare, including technical and privacy risks, lack of patient trust, and the perpetuation of systemic biases, nevertheless it is surely imperative to establish a global baseline for standards in this rapidly evolving area, with a view to ensuring that all patients and communities benefit from certified, safe and equitable healthcare.
It is encouraging to see how individual governments are focusing their efforts on regulating AI in a positive way to tackle major global challenges, of which healthcare is one of the biggest. The impact of HealthAI will need to be assessed over time, and it is clearly a huge mission for a single organisation to undertake. However, for now it is encouraging to see commitment to meaningful global collaboration and cooperation to deliver AI health benefits on an international scale.
Bevan Brittan are the market-leader in the provision of legal, governance and regulatory advisory services to businesses delivering and providing health and care services within the UK and internationally. If you would like to know more about the work we do or to discuss how we can help your business then please get in touch.
Contact Details:
Vincent Buscemi, Commercial Healthcare Partner and Head of Independent Health and Care, Vincent.Buscemi@bevanbrittan.com
Brett Lambe, Technology and IP Partner, Brett.Lambe@bevanbrittan.com
Have some thoughts on the topic? Share them with the community here