Collaborating for Digital Health and Care in Europe

 

With the support of COMFORTage.

 

• “Is AI the magic tool to normalise data?” by Benny Van Bruwaene, BT computing, Belgium
• “How to be sure AI delivers its promise?” by Dr. Tal Patalon, Head of KSM, Maccabi Healthcare Services Research and Innovation Center, Israel
• “The TEF-Health: a new ecosystem for AI implementation” by Prof. Petra Ritter, Charité, TEF-Health Coordinator, Germany
• “Co-creating innovative AI in a European Project around dementia” by Aris Tsitiridis, UBITECH, Greece

🗣️ Speakers

► Session introduction

Andrea Jester, NSBproject s.r.l., Italy

► Is AI the magic tool to normalise data?

Benny Van Bruwaene, BT computing, Belgium

Benny Van Bruwaene took a look at medical coding and AI, and vice versa. He explored four questions, and sought answers to them – first using ChatGPT. His questions covered the challenges of how AI is used in medical coding. Is AI is accurate enough, will medical coding eventually be replaced, and is it likely to be phased out soon? Today, “We need an army of coders to get the work done.” Benny used examples from Belgium to illustrate current coding, including its many millions of euros’ cost to the country’s budget. He ran through various classifications of medical coding – drawing attention to close to 15 types. Manual medical coding is time-consuming and slow, potentially biased, and involves the need for training. Yet AI-based medical coding is still learning how to be unbiased, transparent, legal, ethical, avoid job losses, and to be respectful of privacy. In terms of AI learning, “That’s not so easy in medicine.”

Benny then explored atomic code (a code that cannot be further sub-defined) and its differences in three languages (Dutch, English, and French). Using this approach, he compared and contrasted the terminology used for three conditions: a heart attack, appendicitis, and the use of a specific medicine. His focus was ultimately on how to “do AI in an effective way.” In terms of accuracy, his preference was for SNOMED CT, a clinical vocabulary used in electronic health records, and a “‘snomedised’ approach”. To end, Benny identified three areas in which medical coding might be replaced and/or phased out – even if this takes time and political will.

► How to be sure AI delivers its promise?

Dr. Tal Patalon, Head of KSM, Maccabi Healthcare Services Research and Innovation Center, Israel

Dr. Tal Patalon spoke of the work being done by Maccabi, a major Israeli healthcare provider (health maintenance organisation or health insurer) on AI. As an opening challenge, she suggested provocatively that “digital transformation is far, far in the past; AI … generative AI is the present; the future is quantum computing.” For Tal, AI – which is not just decision-tree algorithms or rules-based algorithms – “is not implemented enough.” Considerable emphasis has to be placed on validation, especially where data from different populations are concerned. “Even if you give a certificate, you need to validate [the algorithm].” She reflected both on methods/procedures for validating algorithms; appropriate places to locate data e.g., the cloud; the use of mobile phones (“cell phones”); and the use of electronic health records. “You can’t have all your data sitting in your basement,” she warned. Overall, she concluded, “It’s about motivation”, rather than the technology. Among Tal’s own key fields of interest in AI are biobanks, genetics, and markers.

► The TEF-Health: a new ecosystem for AI implementation

Prof. Petra Ritter, Charité, TEF-Health Coordinator, Germany

Prof. Petra Ritter’s focus was on trustworthy AI made in Europe. She started by citing the 2024 European AI Act, which she described as the world’s first regulation to enforce trustworthy AI, aimed at creating trustworthiness through both design and default. The legislation’s “grace period” runs out in August 2026. (Among the perceived high-risk areas are medical devices that use AI systems.) How therefore to implement AI in various fields, including in healthcare? Petra introduced the audience to Europe’s aims and claims, and its answer which is a Testing and Experimentation Facility for Health AI and Robotics. It provides technical and scientific support for providers and what are called notified bodies (i.e., certifiers). There is now a single entry point for Europe’s testing and experimentation facilities, including for start-ups and small and medium-sized enterprises. It is headed up by the Charité in Berlin, and provides a “flywheel from lab to market”. The facility has 52 partner organisations from nine Member States around Europe: several – such as from Belgium, France, Germany, Italy, and Portugal – involve many different institutions. One – EBRAINS – is pan-European in orientation. Petra described the facility’s “multitude of services … which can be accessed right now.” She showed how all the services available from the German-led facility function. She came to a close by encouraging people to consider submissions to open calls organised by the facility: “One has been published; another is in the pipeline coming soon …”

► Co-creating innovative AI in a European Project around dementia

Aris Tsitiridis, UBITECH, Greece

Aris Tsitiridis of UBITECH, Greece introduced the audience to COMFORTage. This project is a large-scale project working on 13 pilots to predict, monitor, and personalise the prevention and relief of conditions such as dementia and frailty, with their many risk factors. It is now at the beginning of its implementation stages; community engagement is one of its pillars. “The key is personalisation”, which is able to show, for example, why “what works in one population doesn’t necessarily work in another.”

The project has close to 40 consortium members from 12 European countries, including many university hospitals. EHTEL, and its Norwegian member, NTNU, are among the partners. The project has categorised its pilots into five key clusters on a range of themes, which Aris described in detail (one example is a group of digital innovation hubs and living labs). Extending from COMFORTage’s six objectives, the project’s overall vision is to make healthy ageing achievable for all Europeans by bringing cutting-edge AI to personalised care. One of the project’s main challenges, but also ultimately one of its solutions, is creating an educational AI-based healthcare platform. Overall, the consortium uses a multidisciplinary approach and multimodal systems. These approaches/systems were reflected in the “360 degree angle around the problem” that Aris introduced, and used to explain why such activity is happening now. Among the many approaches that Aris cited was the use of (retrospective) data covering biomarkers, genetics, ‘omics; neuro-mechanics, physiology; and augmented reality, virtual reality, and wearables.

One of the most innovative AI innovations that COMFORTAGE is focusing on is digital twins which provide virtual replicas of patients in order to create proactive healthcare, empower patients, and enable cross-border integration of healthcare. Over its three future years of operation, therefore, COMFORTage is sure to bring interesting AI solutions particularly for clinicians, patients, and carers.

***

This session provided useful insights into some very specific AI-related challenges, such as medical coding. It also opened the door to AI use in politically and geographically diverse settings. They included a European-wide facility for testing and experimentation with AI; a country’s important healthcare provider that uses AI to serve its patients; and more than a dozen AI pilots taking place across the European Union. In its exploration of implementation techniques and approaches, the session provided an interesting preface to the Symposium’s final session on crossing borders in terms of European policy and implementation.