March 2019

How artificial intelligence can revolutionize medicine and pediatrics in the EU

The Situation

Exciting news from the European Commission: Europe wants to stay ahead on artificial intelligence (AI) applications for medicine and aims to become the world-leading region for developing and deploying cutting-edge, ethical and secure AI^[1].

How does this affect children in Europe?

Paediatrics has a specific role in the whole process. Children have an entire life to live and challenges with AI will be harder for them than for adults due to the difficulty in predicting long-term outcomes. Prevention also poses more complex challenges during childhood than adulthood. This also emphasises the need to link different data sources..

AI has the potential to become an essential tool for paediatricians. Itwill help us to anticipate and predict diseases and complications; to get the best from diagnostic images; and, ultimately, to navigate the sea of innovation.

Current Practices

We must admit that Europe has been playing second fiddle to the ”big players” in the US before making the first major steps within this field. Several EU countries have already invested in AI and set it as a high priority in their agenda. The UK NHS has established a long-term plan in which digital medicine and AI play a prominent role. There are, however, certain requirements that should be met before moving further.

First, we must make our medical data available. The adoption of an electronic health record is far from being optimal and a lot of data still exists in hard copy only. Plenty of data is still confined in medical devices connected to patients, unable to be transferred into an electronic health record. Optimistically, no more than 10% of hospital medical devices are connected to a web infrastructure. If we really want to move towards widespread use of AI in medicine, data must be free and available for research, analysis and processing.

Once that has been done, interoperability will be the next main challenge. The existence of large repositories of pooled data requires an agreement upon common essential standards. These already exist, yet interoperability is not seen as  priority and we live in a Babel Tower of data.

Privacy protection has become a challenge to data sharing; fear of incurring in legal issues and the possibility of accessing medical data following patient consent often get mixed up. The General Data Protection Regulation (GDPR) enforcement presents further challenges to data sharing, as people became overzealous out of fear rather than trying to fully understand the rules.

Finally, there is the issue of public-private partnerships. Health professionals cannot play roles that industries know much better. Moreover, health professionals are not educated for entrepreneurship and that is the main reason we are, to a certain extent, afraid of mixing medicine with industrial development. We need technological resources, AI-wise, that we currently do not have in hospitals and medical universities. We are even wary of collaborations with startups, as we do not usually give them much credit.

Our Recommendations

There is only one way to move forward on this issue: expanding the culture of health professionals and not expecting another profession to take the lead role regarding AI implementation in routine clinical practice.

Health professionals are afraid of the unknown and, possibly, of a loss of identity. How can we overcome such a challenge? How can different EU countries connect with each other in order to collaborate and facilitate this process?

The European Commission recommends investing in infrastructures and research.on. Imagine having a unique, large, secure, and substantial way to use data repositories for managing health data, which are continuously under review, analysis and control. Imagine using blockchain to regulate health data transactions. This is not something that depends solely on large and sophisticated infrastructures:, it must begin at a  micro level, with every individual within the health care system. If each of us and each hospital commits to work on their own data structures to make this possible, then we can achieve this. The European Commission has committed to substantial investments to supportthese actions, but nothing will work if we don’t invest in our own culture and challenge ourselves with radical changes in everyday clinical practice.

As formal educational programs of AI in medicine do not exist yet, we should come up with ways and methods that allow us to benefit from this powerful tool. Combining medical culture with data science will be the only way forward.

As Europe pays increasingly more attention to AI, a marvellous opportunity presents itself to paediatricians:: we can work together to suggest and design breakthrough research projects. It’s time to start connecting the dots.

Author:

Alberto E Tozzi, MD
Chief Innovation Officer
Bambino Gesù Children’s Hospital
Rome, Italy

Co-author:

Antonios Loizos, MD
Paediatric resident, Sachsska Children and Youth Hospital
Stockholm South General Hospital Stockholm, Sweden
Communication Officer and webmaster of Young European Academy of Paediatrics
Core member and webmaster of Swedish Junior Association of Paediatrics

Resources:

1. European Commission – Press release. Member States and Commission to work together to boost artificial intelligence “made in Europe”. Brussels, 7 December 2018