The role of AI in public health - future perspectives

The EU AI act was published in the EU Official journal on 12 July 2024 and entered into force across all 27 EU Member States on 1 August 2024. The Act sets out a framework for governance and application of the AI in the EU and foresees to transform the digital health landscape of the EU with high-level objectives: the application of AI in public health possesses the capability to analyse extensive datasets, to forecast disease outbreaks and to tailor treatment plans with targeted intervention.

Benefits vs losing some privacy

Although the use of AI in public health looks very promising and aims to benefit the public, according to Dr. Bertalan Meskó, Medical Futurist, it would be important to help the public understand their role in the AI revolution.

Dr. Bertalan Meskó, Semmelweis Medical School in Budapest: “The right question we have to ask ourselves is how much of our privacy are we willing to give up in exchange for a chance of a healthier life. As long as this decision is ours, ethically we should be fine.” – Read the full interview with Bertalan Meskó

According to him, the key ethical considerations are to help the public understand how losing some of their privacy might benefit them on the long run and how implementing “patient design”, a concept that invites patients as active participants in their own healthcare, could ensure the involvement of patients using AI in healthcare, while fighting against bias.

Concerns, such as data privacy, algorithm bias and transparency requires attention to ensure the responsible adoption of AI.

The EU General Data Protection Regulation (GDPR) safeguards individuals’ privacy rights and practices in public health, however, according to Dr. Meskó, algorithmic bias are a significant concern that need addressing through bias mitigation strategies and representative datasets in order to build trust among healthcare professionals and patients.

“Policymakers should have a crystal clear understanding of the technology to provide the best possible regulations, while healthcare professionals should be involved in the discussions around them as they are the end users of these regulations.”

“Datasets can be misused”

According to Dr. Wannes Van Hoof, researcher at the Belgian research institute Sciensano, the GDPR does not sufficiently provide legal protection for citizens as these datasets can be misused for purposes other than public health. He suggests to develop a framework of collective decision-making based on methods taken from deliberative democracy (a form of democracy in which deliberation is central to the decision-making). This way citizens themselves can establish the rules of the secondary use of their health data.

He points out that while anonymised datasets fall outside GDPR protections, public consensus on AI’s use of health data is essential to align AI applications with citizens’ ethical values.

Indeed, according to a public consultation held by the joint action “Towards the European Health Data Space” (TEHDAS) people also think that they should be considered as a partner in the governance of their health data.

Determining an ethical framework

Dr. Van Hoof believes that the EU framework in the AI Act is doing a good job so far, and empowering local structures, such as ethics commissions, patient review boards, etc. could further facilitate the collaboration between EU policymakers and healthcare professionals.

However, he warns that allowing one group of stakeholders to determine the ethical framework for the development of AI technologies might be dangerous.

Dr. Wannes Van Hoof: “There is a real danger in letting one discipline or one group of stakeholders determine the ethical framework for the development of AI technologies. Everyone has different experiences, expertise, incentives and interests informing their perspective, there is no formula to weigh economic benefits with societal values.” – Read the full interview with Wannes Van Hoof

AI predictive models: working together

AI-driven predictive models have enabled public health authorities to prognosticate and mitigate disease outbreaks and have empowered healthcare professionals in early diagnosis and detection. This underscores the significance of the collaboration among diverse stakeholders, such as EU institutions, academia and the private sector.

In this respect, Dr. Meskó highlights the successful collaboration between the United States’ FDA and the American Medical Association as a great example for EU policymakers and healthcare professionals to follow:

“Policy-makers should have a crystal clear understanding of the technology to provide the best possible regulations; while healthcare professionals should be involved in the discussions around them as they will be the end users of those regulations. A great example the EU should look at is how the United States’ FDA has worked together with the American Medical Association to make it happen.”

Finding balance

Overall, finding the balance between innovation and data privacy concerns requires a multilayered approach in healthcare solutions. Prioritising investments and policies to integrate AI into public health is of paramount importance. We would need to constantly verify and revise EU regulatory frameworks and guidelines in order to ensure adherence to ethical standards, transparency and accountability. The newly established European AI office is there for that.

As AI continues to evolve and integrate into the healthcare system, sustained collaboration in order to navigate its complexities and to unlock its full potential remains crucial. Through fostering innovation while considering ethical challenges and prioritising investments in AI-technologies in public health, the EU can harness its potential to improve health outcomes and patient care for its citizens.