Increasingly long waiting lists, inconsistent diagnoses and tired and overworked doctors, and other healthcare professionals. These are examples of the current unavoidable problems in healthcare, or at least presumed to be inevitable. Since artificial intelligence (AI) is now making its way into healthcare, a large part of the current problems may become a thing of the past. How do Dutch healthcare and AI currently relate to each other? Which opportunities does this create? And what is the limit of the use of 'AI doctors'? Evert Stamhuis, Professor of Law and Innovation at Erasmus School of Law, and Martin Buijsen, Professor of Health Law at Erasmus School of Law, position AI in today’s and tomorrow’s healthcare.
AI is time and cost-saving, does not get tired like doctors, and provides consistency in diagnoses where different doctors may not always be able to do so. Stamhuis also recognises this: “With advanced data-driven technology, such as various learning algorithms, various goals can be achieved in healthcare. However, the variation in our healthcare system must be considered, such as primary care, home nursing and care, hospital nursing, intensive care, and specialised procedures such as diagnosis and therapeutic interventions. In addition, there is a wide range of medical apps accessible to everyone for medical checks, such as Fibricheck for testing your heart rhythm.”
Pros and cons in context
To understand the pros and cons, one must look at the context of an AI application, according to Stamhuis: “For example, the future consultation room will make a different demand on this technology than diagnosing with medical imaging, and this will also vary the consequences for both the healthcare provider and the recipient in terms of ethical and legal expectations. What transparency - such a popular requirement for algorithm use - means is determined, in my conviction, in those concrete contexts.”
“So, we need to look beyond just assistance or replacement of doctors,” Stamhuis continues. “Think about ICU nursing, where there is an expected great need for data-driven support in the next ten years to maintain the level of care.”
Professor of Health Law Martin Buijsen already sees practical examples of this: “Recently, the Santeon hospitals, a group of large non-academic teaching hospitals including the Rotterdam Maasstad hospital, purchased Pacmed Critical-Discharge; AI that assesses the risk of readmission or death of a patient within seven days after discharge from an ICU department. The tool supports ICU staff in making discharge decisions.”
“At the same time, AI has its failure factors,” warns Stamhuis, “once we acknowledge that, it also becomes interesting for lawyers, as responsibility and liability issues come into play. The legal development in that regard is far from complete, even though, for example, EU harmonisation will come for the approval of AI systems and product liability.”
Overenthusiasm
AI is a relatively new and rapidly developing technique, so much must be done to achieve a successful formula. Stamhuis also sees: “A concern is that enthusiasm will prevail over thoughtfulness regarding the promises. Then, various ethical, legal, social, economic, and cultural aspects are neglected. The change in legal relationships in the healthcare chain is important, but equally important is whether access to good care improves or deteriorates. This applies to our country but also globally. When will these expensive technologies be available to healthcare in weaker economies? And are the algorithms suitable for dissemination if they are mainly trained on data from patients in the richer half of the world?”
“Additionally, legislation and regulations are not adequately tailored to data-driven medical research and development. For example, the regulation of personal data heavily relies on individual self-determination over data, which puts pressure on the patient to ask for blanket consent from the patient. That is not a good idea, especially since it threatens to obscure the weight of the importance of a specific study. I also have doubts about voluntariness - a core requirement in the current law,” concludes Stamhuis.
- Professor
- Professor
