Counting Electric Sheep: EU Data Maturity & AI Factories

The healthcare industry is facing significant pressures from a combination of factors, including population aging, changing patient expectations, and shifting lifestyle choices. Without undertaking major structural and transformational changes, European healthcare systems will struggle to remain sustainable in the long run. In other words, the healthcare industry must adapt and evolve to meet the changing needs of patients and society in order to remain viable.

The success of AI technologies, including AI factories, is highly dependent on the quality and availability of data. In the European Union, the sharing of data across borders is crucial for enabling innovative AI initiatives that can bring significant business value and benefit to EU businesses and citizens.

Currently, the sharing of data in the EU is hindered by fragmentation and lack of harmonization across the EU digital single market. This limits the ability of organizations to leverage the full potential of AI technologies, including AI factories. To fully realize the benefits of AI, it is crucial to revise and extend data sharing in the EU, enabling a more integrated and harmonized digital single market. This will help EU businesses and citizens reap the full benefits of AI, including the improved efficiency, accuracy, and insights that can be gained through the use of AI factories.

In November 2023, the European Medicines Agency and the HMA have recognized that AI-powered tools have the potential to revolutionize the way we work by automating tasks, providing instant access to information, and assisting with decision-making the potential benefits of AI models. They have released guidance on technology implementation, suggesting a phased approach, starting with limited use cases in medicine and gradually expanding to more complex applications. (*see below Multi-annual AI workplan 2023-2028) This is particularly important in industries where accuracy and attention to detail are critical, such as clinical research.

In the context of clinical research in the EU, the use of chatbots and LLMs can have a significant impact on the way patient data is collected and analysed. For example, chatbots can be used to collect ePROs and electronic diary data, allowing for more efficient and accurate data collection. Additionally, chatbots can be used to conduct safety follow-up phone calls with patients, ensuring that any potential adverse events are promptly identified and addressed. (*see the EU DCT project) The phased and monitored implementation of these technologies can help ensure that they are integrated into clinical research processes in a way that maximizes their benefits while minimizing any potential risks.

To help visualize the impact for AI in healthcare, McKinsey has mapped six core areas where AI has a direct impact on the patient and three areas of the healthcare value chain that could benefit from further scaling of AI.

AI systems should not only be effective, but also accountable and interpretable, ensuring that they are fair, unbiased, and safe. In the context of the EU, reliable AI has been a focus of the EU's AI regulatory framework, known as the EU Artificial Intelligence Act. This framework aims to promote the development and adoption of trustworthy and ethical AI systems by establishing guidelines for the development, use, and management of AI. Compared to other global regions, the EU has taken a leading role in establishing AI regulation that promotes reliable AI. The EU's approach has focused on ensuring that AI systems are developed and used in a manner that respects fundamental rights, including privacy, non-discrimination, and human dignity. In contrast, other regions have taken different approaches to regulating AI. For example, the United States has focused on promoting AI development and innovation, while China has focused on developing its AI capabilities as part of its broader technological ambitions.

The contrasting approaches of China and the European Union (EU) to the development and deployment of artificial intelligence (AI) systems paint a striking picture of how the aims and practical applications of this technology can shape its impact on society. While China's focus on military uses of AI may undermine stability and trust, the EU's strategy of promoting cooperation highlights a proactive effort to harness the potential of AI for positive, ethical outcomes. This comparison underscores the importance of considering not just the technological capabilities of AI, but also the societal and ethical implications of its development and deployment. (*source: Taylor & Francis Group , Governing artificial intelligence in China and the European Union: Comparing aims and promoting ethical outcomes)

Data Maturity

Data maturity refers to the extent to which an organization has developed its data management practices and capabilities. A data-mature organization has a well-defined data strategy, robust data governance structures, and the ability to extract value from its data assets. The goal of data maturity is to enable organizations to make data-driven decisions, improve efficiency, and drive innovation.

While the indicators used in the European Core Health Indicators (ECHI) have been reviewed in the past in response to scientific developments, changes in data collections, and emerging policy needs, there has not been a systematic and sustainable procedure for this review process. A survey was conducted to assess the availability of data from preferred data sources across 36 countries (EU member states, candidate countries, and EFTA countries). Of the countries contacted, 23 (63%) participated in the survey.

The survey results showed that the availability of data from preferred data sources varied across different chapters of the ECHI. The chapter on demography and socio-economic situation had the highest availability of data, with data available for most indicators from more than 90% of the participating countries. The chapter on health status also had high data availability, with data available for most indicators from more than 90% of the participating countries. However, the survey also identified problems that countries have experienced in incorporating ECHI into their health systems. These problems could include difficulties in collecting and reporting data, lack of resources or capacity to implement the indicators, or challenges in integrating the indicators into existing health information systems. Addressing these problems could help improve the implementation and usefulness of ECHI in monitoring and comparing health outcomes across European countries.

A parallel perspective worth noting is in the most recent Open Data Maturity (ODM) Report by the EU. The ODM assessment focuses on measuring the progress of European countries in promoting and facilitating the availability and reuse of public sector information. It is worth noting that whilst healthcare data is included in the assessment, it is not the sole focus. The ODM assessment may consider factors such as the availability and accessibility of healthcare data, the existence of policies and guidelines for the reuse of healthcare data, and the availability of tools and services for sharing and analysing healthcare data.

The most mature European countries in the field of open data are:

???? France 98%

????Poland 97%

????96% Estonia

????Ukraine most mature EU candidate country

???? Norway most mature EFTA country

AI systems rely on data to learn, make predictions, and generate insights. The ODM assessment provides a framework for evaluating the maturity of open data policies and practices in a particular country or region. By assessing the availability and accessibility of open data, the ODM can help identify areas where improvements are needed, such as increasing the amount of open data available, improving the quality of available data, or making it easier for users to find and access data.

AI Factories

AI factories are designed to optimize the process of developing and deploying AI solutions, providing a centralized and coordinated approach that leverages shared resources, tools, and expertise across the EU. By streamlining the AI development process, AI factories can help organizations accelerate their AI initiatives and extract greater value from their data assets.

One example of a powerful resource that can support AI factories is the LUMI supercomputer, located in Finland. With a computing power of over a quintillion operations per second, LUMI is at the forefront of exascale computing and is central to the EU's AI Factory strategy. As one of the fastest supercomputers in the world, LUMI can provide the computational power needed to train complex AI models and process large datasets, helping AI factories to develop and deploy advanced AI solutions.

In addition to resources like LUMI, the EU has also proposed the creation of an AI Office to oversee the implementation of the AI Act. This office will also play a role in supporting EU startups and SMEs developing AI by providing access to AI-dedicated supercomputers. This initiative aims to foster innovation and support the growth of AI-focused businesses in the EU, further strengthening the region's position as a leader in AI development.

The impact on Healthcare & Life Sciences

European Healthcare Systems

The McKinsey Global Institute studied how automation and AI are likely to affect the future of work and estimates that 15% of current work hours in healthcare are expected to be automated. (source: 麦肯锡 , Transforming healthcare with AI)

The ability of AI to leverage data at scale is a key factor in driving this transformation. By analyzing large amounts of data, AI can automate tasks that are currently done manually, such as data entry, recordkeeping, and certain diagnostic processes. This can free up healthcare professionals to focus on more complex and patient-facing tasks that require human expertise and empathy.

Digitalization of Pharma

The pharmaceutical industry is undergoing a rapid digitalization process, with software being used for a wide range of purposes, from drug development to patient monitoring. This digitalization has the potential to revolutionize the way pharmaceutical companies operate and bring new treatments to market. However, the opportunity must be measured against the risk of fines of 6% of total worldwide annual turnover for infringements related to non-compliance with the prohibition of the AI practices in the EU (*source: Simmons & Simmons, EU Artificial Intelligence Act).

Impact on MedTech

There are significant implications for the medical technology industry, medical devices, including medical device software and in vitro diagnostic medical devices, are explicitly regulated. This will impact how products are designed, tested, and marketed. The new MDR places a greater emphasis on clinical evidence and post-market surveillance and introduces stricter requirements for high-risk devices. (*CE marking of conformity)

Concerns for Digital Health Stakeholders

Many stakeholders in the digital health space may be concerned as their technologies may be classified as low-risk or high-risk AI systems, even if they have previously escaped CE marking requirements for medical devices. This could have implications for how these technologies are regulated and how they can be used in healthcare settings.

As we consider the impact of AI on Healthcare and Life Sciences, it becomes clear that data is a critical driver of AI's transformative potential. AI systems rely on data to learn, improve, and generate insights that can enhance patient care, streamline research, and unlock new discoveries. However, not just any data will do - the key to unlocking AI's full potential lies in data at scale.

Much like the androids in Philip K. Dick's classic novel "Do Androids Dream of Electric Sheep," which counted electric sheep to induce sleep, AI systems rely on vast amounts of data to function effectively. Just as the androids needed a sufficient number of electric sheep to achieve their desired state, AI systems need a sufficient amount of data to achieve their full potential.

In the context of Healthcare and Life Sciences, this means accessing large, diverse, and high-quality datasets that can power AI models and algorithms. By leveraging data at scale, AI can uncover patterns and insights that would be impossible for humans to discern, leading to more accurate diagnoses, personalized treatments, and faster drug development.

However, the pursuit of data at scale must be balanced with considerations of privacy, security, and ethical use. As we strive to harness the power of AI, we must ensure that the data we use is not only abundant but also trustworthy, responsibly sourced, and respectful of individuals' rights. Only by counting our data "sheep" carefully and ethically can we truly unlock the promise of AI in improving health outcomes and transforming healthcare for the better.