Artificial Intelligence

Erasmus MC and TU Delft open first healthcare AI ethics lab

Erasmus MC will open the world’s first ethics lab for artificial intelligence (AI) in healthcare. In this lab, nurses, doctors, data scientists, data engineers and ethicists work together to ensure that artificial intelligence can be employed safely in patient care. ‘Would doctors discontinue treatment because the computer says so?’

4 likes
Reading time 3 min
EthiekLab_3-1920×1080

The new AI ethics lab aims to ensure that artificial intelligence can be used safely in patient care. ‘Because of the coronavirus crisis, we are facing shortages of nurses in hospitals and the entire healthcare chain’, explains Michiel van Genderen, internist-intensivist at Erasmus MC. ‘The ICU department concluded that we had to adapt. That is why we started the DataHub early last year to develop AI models.’

REAiHL

The ethics lab is named Responsible and Ethical AI in Healthcare Lab (REAiHL) and is an initiative by Michel van Genderen, an internist intensivist at Erasmus MC. Experts from the convergence of Erasmus MC, TU Delft and Erasmus University are involved. Software company SAS is making the initiative possible by sponsoring PhD students, enabling a total of five PhD students to work in REAiHL. In working with hospitals, SAS focuses on developing and implementing AI. They consider it important to embed ethics in the technology as much as possible. The location of the ethics lab is Erasmus MC’s DataHub.

 

More time

AI can use data and computer models to predict patient outcomes, making the healthcare system more efficient. It is one of the possible solutions to the ever-increasing demand for care and the shortage of personnel, Van Genderen believes. ‘This enables the medical staff to devote more time to the patient, which ultimately ensures that staff does not leave, but rather remain in the healthcare field: less workload, better quality.’

‘Currently, less than two per cent of all AI studies make it into clinical practice’

Erasmus MC wants to embrace the developments of AI, but also finds it important that implementations of AI are explainable and comply with all ethical principles, Van Genderen says. ‘In the next three to five years, we hope that the computer can help us decide whether further treatment in the ICU makes sense or not for a particular patient. Currently, medical specialists make these decisions in patient consultations. Sometimes we feel that we lack the latest insights, so we want to know if data and AI can help us decide.’

Key question

Would doctors discontinue treatment because the computer says so? That is the key question, according to Van Genderen. ‘The model advises, but the healthcare worker remains in charge.’

‘We hope the ethics lab will bring peace of mind.’

Co-initiator Diederik Gommers notices a great need in practice for AI ethical principles. ‘Currently, less than two per cent of all AI studies make it into clinical practice, partly because ethics are lacking.’ Studies show that 77 per cent of ICU nurses are ready for AI, but there are also many questions and uncertainty. ‘This is because, for example, there is a lack of clarity about responsibility’, Gommers knows. ‘We hope the ethics lab will bring some peace of mind.’

Ethical principles

The World Health Organization (WHO) has established six ethical principles for AI in healthcare. One of the most important, according to Van Genderen, is equality. ‘AI models are developed based on data, and in those datasets, you can often find inequality, for example in skin colour. It is important to be aware of this and work toward making AI models apply equally to everyone. That is why we are honoured that experts from TU Delft, such as Jeroen van den Hoven, Stefan Buijsman and Jacobien Oosterhoff, are part of the AI ethics lab.’

Jeroen van den Hoven, who helped the WHO draft the principles, says: ‘AI use in healthcare will really take off in the coming years. We are very pleased that together with our colleagues at Erasmus MC and with SAS supporting a group of PhD students, we can now really start working on the WHO’s ethical principles for Responsible AI in the clinic.’

Soon, nurses, doctors, data scientists, data engineers and ethicists will work together in the ethics lab. ‘In the coming period, we want to approach even more parties to see if they want to join as well’, says Van Genderen. ‘We will start with the patient council.’

Recognition

The National Innovation Center for Artificial Intelligence (ICAI) has recognized the REAiHL as an official lab. An ICAI lab is a research collaboration between industry, government or nonprofit partners and knowledge institutes. ICAI labs must meet requirements for data, expertise and capacity. They are expected to operationalize outcomes for the real world.

Pictured are (from left to right) Phaedra Kortekaas (SAS), Diederik Gommers (Erasmus MC), Antonie Berkel (SAS), Reggie Townsend (SAS), Jacobien Oosterhoff (TU Delft), Michel van Genderen (Erasmus MC), Stefan Buijsman (TU Delft), Jeroen van de Hoven (TU Delft).

Also read