A new whitepaper explores reasons for the gap between research and practice in healthcare. Human factors, technical challenges and infrastructural aspects of artificial intelligence (AI) in the health domain are discussed to create a starting point for researchers, practitioners and policy makers.
The whitepaper is an initiative by VU Campus Center for AI & Health. Three institutes, Vrije Universiteit Amsterdam, Amsterdam UMC and Amsterdam University of Applied Sciences collaborate in multidisciplinary networks and consortia to address AI challenges from all relevant angles with people each bringing their expertise to the table. The paper addresses the underlying mechanisms of the existing implementation challenges. Ultimately, such efforts are essential to develop innovative, human-centred AI solutions with real-life positive outcomes for patients, relatives, healthcare workers and society at large.
Potential to provide relief in healthcare
The societal need for innovation in health and healthcare is clear. The pressure on healthcare is rising, and AI can relieve the pressure on healthcare professionals with decision support and taking over administrative tasks. ''The population is ageing, with a record number of cancer diagnoses and a significant rise in mental health disorders. To meet the increased care demand in 2050 estimates show almost one in three people of working age in the Netherlands need to be employed in healthcare. AI has the potential to provide relief on many levels'', Mark Hoogendoorn explains, AI scientist at VU Amsterdam and Amsterdam UMC, collaborating under the umbrella of VU Campus Center for AI and Health.
Success depends on human factors
The successful integration of AI in healthcare depends on a few key human factors: co-creation, AI knowledge and skills among healthcare professionals, and user accountability and acceptance of AI models. It is essential to address all these factors as they directly impact the acceptance, usability, and effectiveness of AI in healthcare, ultimately influencing patient outcomes and quality of care. The researchers see AI as a complementary tool, rather than a replacement for healthcare professionals. ''In breast cancer screening, for example, it has been shown that combining an AI system with the expert knowledge of a radiologist leads to improved accuracy as compared to either the individual radiologist or a standalone AI system'', says Hoogendoorn.
AI reliability beyond metrics
But the reliability of AI models is challenged by the complexity, size, and completeness of datasets. Once developed, AI models may not generalise well. ''For me as a healthcare professional it is imminent that the AI model is trustworthy, robust and that I am able to understand why the model provides a certain output'', says Edwin Geleijn, physiotherapist at Amsterdam UMC. Also, evaluations in real-world settings remain essential.
Challenges in AI lifecycle
After the development and evaluation of a model, there are challenges that arise at the deployment and monitoring phase of an AI lifecycle. Such as legal issues around the use of data and AI in clinical practice, data availability and sharing between institutions, technical implementation and integration in clinical workflow. Martijn Schut, AI scientist at Amsterdam UMC, adds: “The time and effort needed to go from development and research to implementation of healthcare AI applications is generally underestimated, but doing this accurately and meticulously is essential to its effective and reliable use”.
Collaborating for reliable AI in healthcare
Promising directions include not only technical innovations for AI development and data management, but also organisational innovations to improve the design process, the AI training of medical professionals, and the monitoring of AI in practical settings. Therefore, professionals need to dive in the challenges and perspectives of human, technological and infrastructural factors. The researchers look forward to exploring these innovations with the communities involved in AI for healthcare beyond Amsterdam.
This article was written by VU.