Academic Alliance Radboudumc – Maastricht UMC+ is developing an artificial intelligence application in private cooperation with companies, which enables the visualization of DNA, brain signals or hearing, for example. In this way, patients should gain a better understanding of their personal health status. The PROSPER InnovationLab consortium, which includes a total of 27 partners led by ENT doctor Marc van Hoof of the Radboud University Medical Center, will receive 3.5 million euros from AiNed to develop this medical visualization.
According to the researchers involved, communication about medical information between healthcare providers and patients will change completely in the next few years. Instead of a doctor having to explain black and white scans step by step, in the future artificial intelligence will produce insightful images of an individual patient’s anatomy and abnormalities. Trained LLMs (large language models) like ChatGPT can explain and even visualize medical information in an understandable way. This enables easy interpretation of complex data sources such as DNA analyses, brain signals and hearing tests. Van Hoof expects the AI application to lead to new insights, better diagnostics and a better understanding of patients’ health.
AiNed is a Dutch growth fund that is committed to developing artificial intelligence both economically and socially. It announced in mid-October that it had selected four innovation labs and five learning communities to support the grant. Led by Radboud UMC, Radboud University and Maastricht University Medical Center (MUMC+), the PROSPER InnovationLab is developing a platform for secure information exchange between hospitals using artificial intelligence to improve patient-healthcare provider communication through visual anatomical views,” AINed said at the time.
Obstacles due to privacy legislation
Privacy legislation, especially GDPR, is currently still an obstacle to the development of artificial intelligence. Training AI applications requires a lot of data, but developers can’t just use traceable patient data. According to Maastricht UMC+ researcher Joost Stultiens, PROSPER InnovationLabs claims to have come up with a solution to this. “We’re going to build very large synthetic datasets of real data, but made completely anonymous and therefore no longer traceable.”
This approach should enable data exchange between Dutch hospitals on a large scale, while complying with data protection legislation. The aforementioned datasets, which are no longer known from which patient they originated, should enable developers to quickly launch an AI project without years of approvals and contract negotiations. According to Ronald Pennings, professor of ENT at the Radboud University Medical Center, this scale is only possible thanks to the partnerships formed between Dutch hospitals. “We want to collaborate to enable real innovations that improve care.”
A breakthrough in artificial intelligence training
The above method allows patients to manage their data online through a dynamic consent procedure, depending on the nature and sensitivity of the data. According to the startups, this combination of anonymized health data and dynamic patient consent could provide a breakthrough in obtaining the necessary data for AI training in healthcare. This means that traceable data no longer needs to be shared with commercial entities, and patients can decide for themselves who can use their data and for what purpose.
Van Hoof and his team are starting with this approach to treat ENT diseases, but expect rapid translation to other areas of medicine. “We demonstrate this by combining various medical disciplines, from neurosurgery to breast cancer diagnosis at home,” said the otolaryngologist. “This not only saves time and energy, but also reduces the need to collect large amounts of patient data for each application, strengthening both healthcare and the Netherlands’ commercial position in AI health technology.”
The PROSPER InnovationLab consortium includes: Radboudumc, Maastricht UMC+, UMCG, TNO, Superconnectors, Enatom, SURF, WSK Medical, IDS International BV, Quantitas Solutions, Earhelp, Zuyderland Medical Center, Catharina Hospital, Deventer Hospital, Elisabeth-Tweestala Clinics, Rijnstate, Bernhoven Hospital, Elkerliek Hospital, Treant Hospital Care, AI hub Oost NL, AI hub Brightlands, Brainport, Screenpoint medical, Dutch Head Neck Society (NWHHT) and Eyehelp.