Guideline for high-quality diagnostic and prognostic applications of AI in healthcare
This guideline provides a description of what the work field considers good professional conduct in the development, testing and implementation of an Artificial Intelligence Prediction Algorithm (AIPA) in the medical sector, including public healthcare.
The requirements and recommendations in the guideline are addressed directly to the developers and testers of the AIPA, the manufacturer of the software that incorporates the AIPA and the care organisation that implements this software in its organisation. The intended audience includes manufacturers of devices that include an AIPA, researchers developing and testing an AIPA, care organisations and care providers who wish to purchase and use such devices and authorities that help to determine the quality, deployability and reimbursement of the AIPA. The guideline describes what care providers, citizens and patients, insurance companies and policy-makers (such as the National Health Care Institute and the Dutch Healthcare Authority) can expect from an AIPA developer or manufacturer when they purchase such medical devices, use these devices or have such devices used on them.
By and for the healthcare field
This guideline has been developed by and for the field. Many participants contributed to the development with passion and attention.
Maarten van Smeden, Carl Moons, Lotty Hooft (fase 1 t/m 3)
Ilse Kant, Hine van Os, Niels Chavannes (fase 4 t/m 6)
Working group members:
Fase 1: Paul Agra, Amy Eikelenboom, Andre Dekker, Christian van Ginkel, Saskia Haitjema, Martine de Vries.
Fase 2: Gabrielle Davelaar, Desy Kakjay, Evangelos Kanoulas, Kicky van Leeuwen, Joran Lokkerbol, Daniel Oberski.
Fase 3: Huib Burger, Daan van den Donk, Vincent Stirler, Wouter Veldhuis, Bart-Jan Verhoeff.
Fase 4: Giovanni Cina, Marcel Hilgersom, Maurits Kaptein, Floor van Leeuwen, Martijn van der Meulen, Egge van der Poel.
Fase 5: René Drost, Sade Faneyte, Leo Hovestadt, Teus Kappen, Ewout Steyerberg, Jonas Teuwen, René Verhaart.
Fase 6: Nynke Breimer, Bart Geerts, Anne de Hond, Lysette Meuleman, Karen Wiegant, Laure Wynants.
Supervised by Maarten van Smeden and Ilse Kant, with support of Roy Tomeij, Pieter Boone and Rachel Peeters.
Annemarie van ’t Veen, Charlotte Brouwer, Daniel Vijlbrief, Elise Quik, Jan-Jaap Visser, Jan-Kees van Wijnen, Jan-Willem Wasman, Jean-Paul Kleijnen, Joris van Dijk, Leon Doorn, Lieke Poot, Maaike van Mourik, Mark Scheper, Martin van Buuren, Merel Huisman, Richard Bartels, Rimmert Brandsma, Rob Tolboom, Roel Streefkerk, Roel van Est, Wouter Bulten
Practical test participants:
RetCAD, Thirona - Mark van Grinsven
HUME, Mentech – Reon Smits en Erwin Meinders
U-Prevent, Ortec – John Jacobs
Risicotaxatie agressie in de psychiatrie, UMCU – Karin Hagoort
Covid-19 severity score, Maasstad ziekenhuis – Sade Faneyte
Feedback has been processed by the editorial board consisting of the authors under the guidance of Ylja Remmits and Alexander Boer (KPMG Trusted Analytics).
Action team members:
Jan Jaap Baalbergen (NFU), Robert Geertsma (RIVM, Dennis Japink (ZN), Carl Moons (UCMU), Rozemarijn Pennings (InEen), Marlies Schijven (Amsterdam UMC), Jaap Schrieke (GGZ Nederland), Inge Steinbuch (ActiZ), Jos Schimmelpennink (Nederlandse Vereniging van Ziekenhuizen), Stefan Visscher (Federatie Medisch Specialisten) en Claire de Monte (Ministerie van VWS).
Questions or remarks?
The development of the guideline was one of the action lines of the program 'Waardevolle AI voor gezondheid' of the Dutch ministry of VWS. The coordination was in the hands of Rosalie van Oostrom. With the delivery of this guideline the action line and the program have ended.
Questions and comments can be shared on the LinkedIn page.