Abstract
Introduction
Introduction: The use of artificial intelligence (AI), including autonomous systems
and robotics, is expanding across sectors, including healthcare [1]. Initially, the
development and safety assurance of AI was technology centric, focusing narrowly
on the algorithm, and how it for example compared with expert clinicians. However,
it has become increasingly apparent that a human-centred approach is
necessary, which focuses on how the AI technology will interact with the wider
sociotechnical work system to enhance overall system performance, safety and
human wellbeing [1]. To highlight how a human-centred approach for AI
technologies development and assurance may be completed, this project aims to
develop a contextualised human-centred AI framework, based on previously
completed industrial demonstrators and the wider literature. While this will be a
cross-sector framework, to showcase its benefit to healthcare, the framework will
be mapped against an AI technology developed for the healthcare sector.
Methods
Firstly, 22 industrial demonstrates completed as part of the Assuring Autonomy
International Programme (AAIP) were examined to provide context and highlight
the types of work system interactions studied, alongside the approaches and
methods used and the key results [2,3]. Secondly, a scoping review of existing
human-centred frameworks will be completed to highlight key principles that
should be of focus. The results from the reviews will then be used to develop the
contextualised human-centred AI framework, which will be mapped against an AI
technology, Dora, a voice-based clinical AI assistant and medical device used in
the NHS by thousands of patients weekly, to highlight its usefulness in healthcare
[4].
Results
The review of industrial demonstrators highlighted that the most common
interaction studied was between the AI technology and its user, both on its own and
within a broader environmental context. One demonstrator looked at all work
system elements, and several focused on the interaction between the AI
technology and the external environment, including regulation [3]. The scoping
review is currently in progress, but this alongside the finished contextualised
human-centred AI framework will be presented at the conference.
Conclusions
This project aims to exemplify how a human-centred approach can be taken for
the development and assurance of AI technology, across sectors, including
healthcare. By applying the two review methods, the framework will showcase the
high-level principles suggested in the literature alongside the interactions
considered within the industrial context.
Introduction: The use of artificial intelligence (AI), including autonomous systems
and robotics, is expanding across sectors, including healthcare [1]. Initially, the
development and safety assurance of AI was technology centric, focusing narrowly
on the algorithm, and how it for example compared with expert clinicians. However,
it has become increasingly apparent that a human-centred approach is
necessary, which focuses on how the AI technology will interact with the wider
sociotechnical work system to enhance overall system performance, safety and
human wellbeing [1]. To highlight how a human-centred approach for AI
technologies development and assurance may be completed, this project aims to
develop a contextualised human-centred AI framework, based on previously
completed industrial demonstrators and the wider literature. While this will be a
cross-sector framework, to showcase its benefit to healthcare, the framework will
be mapped against an AI technology developed for the healthcare sector.
Methods
Firstly, 22 industrial demonstrates completed as part of the Assuring Autonomy
International Programme (AAIP) were examined to provide context and highlight
the types of work system interactions studied, alongside the approaches and
methods used and the key results [2,3]. Secondly, a scoping review of existing
human-centred frameworks will be completed to highlight key principles that
should be of focus. The results from the reviews will then be used to develop the
contextualised human-centred AI framework, which will be mapped against an AI
technology, Dora, a voice-based clinical AI assistant and medical device used in
the NHS by thousands of patients weekly, to highlight its usefulness in healthcare
[4].
Results
The review of industrial demonstrators highlighted that the most common
interaction studied was between the AI technology and its user, both on its own and
within a broader environmental context. One demonstrator looked at all work
system elements, and several focused on the interaction between the AI
technology and the external environment, including regulation [3]. The scoping
review is currently in progress, but this alongside the finished contextualised
human-centred AI framework will be presented at the conference.
Conclusions
This project aims to exemplify how a human-centred approach can be taken for
the development and assurance of AI technology, across sectors, including
healthcare. By applying the two review methods, the framework will showcase the
high-level principles suggested in the literature alongside the interactions
considered within the industrial context.
| Original language | English |
|---|---|
| Title of host publication | 8th International Triennial Conference on Healthcare Systems Ergonomics and Patient Safety (HEPS) 2025 |
| Publication status | Published - 21 May 2025 |
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver