AI Deployment Journey in Healthcare: Governance, Design, and Adoption

In the Netherlands, healthcare costs are rapidly increasing just like the shortage of staff. At this moment, Dutch hospitals are already having trouble delivering the care their patients need and the outlook will only get worse in the near future. Due to the growing and aging population, the World Health Organization predicts a large shortage in healthcare workers worldwide of 10 million by 2030.

Artificial Intelligence (AI) is considered to play a crucial role in solving the problem of staff shortage by increasing workflow efficiency and simultaneously providing a better quality of care. AI can be used in the healthcare context for disease diagnosis, precision medicine, predictive modeling of patients at risk and more efficient and effective care processes using techniques such as natural language processing, data analytics, and machine learning. However, despite the great potential of AI to be used in healthcare domains, its actual deployment continues to be a topic of debate.

The main issues of applying AI in healthcare practice concern the potential errors and patient harm, algorithmic bias caused by bad data and the application outside the scope, lack of transparency, persistent trust issues, the difficulty of deployment in the routine clinical care, privacy issues of medical records, liability with AI, and the risk of increased health inequalities. Aim of this project is to develop an integral system of practices, processes, and technological tools to support healthcare organizations’ deployment of AI technologies.

Project objectives

Effective AI deployment requires that relevant stakeholders (e.g., end-users, management, and inspectorate) are involved in the governance and design of such systems from the beginning, highlighting the importance of explainability in AI deployment.

The overall outcome of this project will be an integrated framework that can be offered to experts in a manner that supports their decision making and at various levels of granularity, with the reflective perspectives on explainability and can therefore contribute to trust. This will be achieved through the advancement, application, evaluation, and integration of data science and design science techniques, as well as quantitative and qualitative methodologies in the context of AI deployment.

As the AI deployment in healthcare is regarded as a journey that covers governance, design and adoption, these three components are covered in separate work packages:

• AI governance and explainability to establish novel frameworks, policies, and standards that ensure effective deployment of AI in healthcare;
• AI design and deployment to develop innovative AI algorithm, interfaces and reporting structures tailored to address the unique challenges in engaging clinicians to give feedback to an AI model;
• the role of trust in AI adoption in order to identify and build calibrated trust aiming to increase acceptance and deployment of AI systems among healthcare professionals, patients, and other stakeholders.

As use cases, the deployment of AI tools in the clinical workflows in this project will cover head-and-neck tumor treatments, natural language processing for clinical trial patient selection, explainable AI for dose-guided radiotherapy and AI tools for detection and process monitoring of bone lesions.

Potential impact

Primary objective of this ambitious initiative is to revolutionize the management process of AI in healthcare organizations and improve the collaboration between AI systems and humans in clinical decision-making. It primarily focuses on leveraging automatic diagnosis from radiology images to achieve these goals. Traditionally, healthcare decisions, particularly in tumor detection and treatment, have heavily relied on the experience and knowledge of experts. However, this project recognizes the potential of utilizing AI-driven approaches to make well-informed decisions. By implementing a cultural shift towards underpinning decisions by data-driven methodologies, healthcare organizations can significantly enhance the efficiency and effectiveness of processes related to tumor detection, reporting, and formulating cancer treatments.

In addition to improving industrial and economic aspects, this project also emphasizes ecological sustainability. By optimizing the deployment of AI in healthcare, the project aims to reallocate healthcare manpower and minimize long patient waiting times. This optimized AI deployment journey will ultimately result in higher service availability for patients at reduced costs. The integration of accurate, data-driven medical processes and decisions will significantly benefit the healthcare industry and patients. Importantly, the methods developed in this project have the potential to apply AI deployment to other complex diseases that require close control and monitoring. This versatility makes the project valuable beyond the current use cases and healthcare, benefiting various domains where AI plays a crucial role but needs careful management.

Overall, this project on AI deployment will contribute to society as a whole by promoting healthcare safety at lower costs and offering more sustainable technological solutions. It represents a significant step towards achieving a well-aligned and better-deployed AI lifecycle management process, ensuring that AI and human professionals work together harmoniously to provide optimal care and decision-making capabilities in healthcare and beyond.

Duration

The project will run from January 2024 – December 2027.

Multidisciplinary project consortium

In this project, a team of scientists from Tilburg University, as well as ETZ Hospital, Maastro Clinic, together with the support from Dutch Association of Hospitals (NVZ), will collaboratively address the AI deployment research questions and co-develop the AI Deployment Framework that covers the critical elements of governance, design, and adoption for the healthcare industry.

The team is dedicated to establishing connections between digital interfaces, social dynamics, economic factors, and management practices and comprises of experts from diverse fields such as Artificial Intelligence, Data Science, Informatics, Health Science, Economics and Management, Social Sciences, and Humanities.

Prof. dr. Carol Ou (principal investigator, TiSEM) is currently Full Professor of Digital Transformation and Information Management and the Head of the Management Department at Tilburg School of Economics and Management (TiSEM), Tilburg University. Carol is a distinguished member of the Association of Information Systems and serves as a Senior Editor for three top journals in the field of information. Carol's current research interests now focus on digital transformation, primarily investigating how healthcare, military, governmental and commercial organizations are transformed by new technologies like AI.

Prof. dr. Maria Jacobs (work package leader, TiSEM) is the endowed chair on innovation implementation in health care at the Management Department of TiSEM and CEO of the Board of Directors at Maastro. Maria has a drive for innovation, is educated in business administration, HR and management, and has a demonstrated history of working in the hospital & health care industry for almost 30 years. Prof. dr. Maria Jacobs works on research and education in the field of innovation implementation in health care and has extensive experience in coordinating complex innovative projects, including the establishment of a new Proton Therapy centre, the building of a satellite centre and numerous hospital-level projects.

Prof. dr. Marie Postma (work package leader, TSHD) is currently Head of Department and Professor at the Cognitive Science and Artificial Intelligence Department. Marie is Research Leader of the Computational Brain and Behaviour Group. In her research she focuses on modeling human cognitive states using behavioral and psychophysiological data for the purposes of human-machine teaming and adaptive human-AI interfaces.

Dr. Sharon Ong (work package leader, TSHD) is an Assistant Professor at the Cognitive Science and Artificial Intelligence Department. She is the Principal Investigator of the Deep Learning for Medical Image Data Group in the Deep Learning for Perception Unit. Her current research interests are in medical image analysis and implementation of diagnostic imaging tools into practice.

Prof. dr. Anne Marie Weggelaar (work package leader, TSB) has led over 40 research projects, including a large EU funded project on the uptake of AI technology in healthcare (BigMedilytics). Her position at Tilburg University is funded by the 27 teaching hospitals in the Netherlands (STZ), who benefit from this research project. Researchers from the SSH Digital Health and Wellbeing team. This multidisciplinary research team aims to study the interaction between individuals, teams, organizations, institutions and/or society at large and digital technologies within the field of health and wellbeing, using an interdisciplinary approach. It includes the following researchers: Dr. Roshnee Ossewaarde, Dr. Gubing Wang, Mr. dr. Charlotte Zegveld, Dr. Gert Meyers, Dr. Matti Vuorre, Dr. Seyed Mostafa Kia and Dr. Ulrich Laitenberger.

Prof. dr. Antoinette de Bont (TSB) joined Tilburg School of Social & Behavioral Sciences (TSB) in 2021 as dean. Prior to TSB she was vice dean education at the Erasmus School of Health Policy and Management (ESHPM) and professor in sociology of innovation. Her research focuses on digital health care. Together with healthcare providers, policy makers, engineers and businesses, she studied the development of big data in healthcare in Europe. She is a graduate of Health Sciences at Maastricht University, where she majored in governance studies and technology studies.  

Dr. Aswin van Oijen (TiSEM) is an associate professor of strategy and organization at the Management Department. He studies the key challenges and opportunities that organizations are confronted with and the strategies that they subsequently design and implement to ensure their long-term sustainability.

Drs. Luca Heising (TiSEM) is a PhD candidate in Maastricht University, co-supervised by Prof. dr. Maria Jacobs and Prof. dr. Carol Ou. Luca obtained her Master’s degree in Cognitive Science and Artificial Intelligence at Tilburg University. Her PhD research focuses on how explainable artificial intelligence can bridge the gap between AI development and AI implementation in healthcare.

Dr. Gorkem Saygili (TSHD) is an Assistant Professor at the Cognitive Science and Artificial Intelligence Department. He received his PhD in Computer Science from the Delft University of Technology. His current research interests are in medical data analysis, medical image registration, confidence estimation, and continual learning.

Maastro is the only radiotherapy center in the Netherlands awarded by the Ministry of Healthcare with the status of being a top specialist center (TZO) providing radiation oncology treatment to patients with cancer. The status is assigned because Maastro has proven to be able to conduct high level research, to implement the results of this research into clinical practice. One of the spearheads in the strategy of Maastro, taking into account the TZO-mission to make healthcare impact, is to develop knowledge on how to successfully implement more AI in healthcare, in order to make healthcare more sustainable. Maastro is collaborating with many other hospitals in the Taskforce Innovation Implementation Radiotherapy in the Netherlands on this topic with the goal to increase the societal impact of research findings. On behalf of Maastro, the following experts are involved in the project:

• Dr. ir. Wouter van Elmpt, Assistant Professor and head of the physics innovation implementation team in Maastro.
• Drs. Rachelle Swart, PhD candidate, with her PhD research focusing on the implementation part of AI, using validated implementation frameworks and strategies.
• Dr. Paul Cremers, manager of research and project leader of the TZO project at Maastro.

Because of the combination of clinical leadership and scientific experience of the above-mentioned participants in this study, Maastro will be a use-case with all conditions for successful implementation secured.

Elisabeth-TweeSteden Ziekenhuis (ETZ) is a top clinical hospital and healthcare provider in the south of the Netherlands. It is one of the Samenwerkende Topklinische Ziekenhuizen (STZ) in the Netherlands. ETZ will collaborate on implementing an AI model for osteolytic bone lesion detection and progress monitoring. On behalf of ETZ, the following experts are involved in the project:

• Dr. Gerlof Bosma, radiologist and Dean at ETZ Hospital
• John Creusen MSc, medical physicist and member of the AI Governance Board at ETZ Hospital

Other members of the team include Dr. Bastiaan Steunenberg, radiologist, and three radiology residents, Dr. Thijs van Oudheusden, Dr. Bas Oei and Dr. Rik Kint.

This project is funded by Tilburg University’s Digital Sciences for Society program: