Digital pathology and service design
Posted 6th October 2022 by Nick Noakes
Digital health solutions in general have a poor track record of being sustained once they are implemented, often resulting in abandonment of the technology. Current digital pathology and artificial intelligence (AI) deployment strategies are generally too IT-focused – the technology being the focus, rather than the people and environment into which they are being deployed. Furthermore, a focus on technology alone will also hamper the effective integration of AI tools into the diagnostic workflow.
This article was written by Samar Betmouni, Consultant Neuropathologist, Sheffield Teaching Hospitals NHS FT. It was originally published in The July 2022 Bulletin of the Royal College of Pathologists. It is republished here with their kind permission.
The panel discussion held during the 8th Digital Pathology & AI Congress in London (1–2 December 2021) explored how good design practice principles can contribute to deploying and sustaining digital pathology, along with the effective integration of artificial intelligence into the diagnostic workflow.
Looking ahead to how AI can best be deployed, we will need to ensure effective mechanisms to appraise the technology, ensure the avoidance of bias and address ethical and medicolegal issues that will arise with the use of AI as a diagnostic tool. We will also need to have appropriate education and training resources to promote user and stakeholder confidence in how the AI operates in their speciality and healthcare as a whole.
What role can service design play in digital
Professor Cain reminded us of a Cedric Price quote: ‘Technology is the answer but what was the question?’
She suggested that the challenge of designing a digital pathology service is knowing what the requirements of digital pathology deployment are. It is important to first consider the human factors that are an intrinsic part of deploying innovation to start with, rather than focusing on the technology and facilities. In this respect, it is useful to look outside of healthcare to see what we can learn from other sectors that are deploying technologies that can impact people in the workplace. Fundamentally, services are centred on people and experience and so developments need to be co-creative – created from the bottom-up, holistic, transdisciplinary and integrative – bringing together different disciplines to address a complex, systemic problem.
In pathology, we are familiar with working in multidisciplinary teams. A transdisciplinary team provides a platform for incorporating new ways of thinking by working with colleagues from disciplines outside of those we conventionally work within digital pathology, such as pathologists, biomedical scientists, computational scientists and IT support.
By widening the make-up of digital pathology deployment teams to include designers, we would hope to provide a new understanding of the relationship between science and society to transcend disciplinary views and enable critical reflection to create new knowledge.
Incorporating design specialists into digital pathology and AI deployment projects brings with it the application of design tools (such as customer journey maps, personas, stakeholder maps, and service blueprints) and so allows exploration and reframing of challenges. This provides opportunities to investigate the perceived benefits or disadvantages of a technology. Designers will ask focused questions that look at the deployment holistically. They can use the ‘How Might We?’ (HMW) question to reframe challenges. In the case of digital pathology, the HMW could be:
- HMW re-imagine and co-create meaning and value in digital pathology for all stakeholders.
- HMW use technology to enhance human experiences within digital pathology.
For each of these questions, Professor Cain highlighted the need to know if digital pathology must look outside of its own discipline and, if so, which disciplines should be involved.
Understanding technology and innovation
Dr Jun reminded us that the digital revolution has huge potential in healthcare. If done well, it can support clinicians in decision-making, generate predictions, improve inefficiency in care management and radically transform the way care is accessed and provided. However, it is also clear that innovations do not always work, as seen in past examples including telehealth and care technologies, which were purchased by local authorities but left unused and even unpacked.
This arises because of a primary focus on the technology rather than understanding deployment in the context of the overall system. A mismatch also often exists between existing systems and new technologies, which further compounds the challenges such as poor interoperability, usability or consideration of implementation within the system. This mismatch can lead to the disruption of care delivery and even contribute to stress and burnout in clinicians or harm to patients.
There is a need to move from the technology-focused view to a systems perspective, especially for AI deployment. The former tends to focus mainly on data quality, bias in data, algorithm accuracy and human versus machine performance. On the other hand, a systems perspective provides a more holistic approach involving multiple stakeholders. It considers the following issues highlighted in a white paper on human factors and ergonomics in healthcare AI:
- how AI can help people to be better aware of their situation
- how people can appropriately interpret automated decisions made by AI
- how AI and people can work as a team
- how trust can be built between AI and people and what ethical concerns can be raised.
Importantly, the system perspective asks basic questions that provide the basis for effective technology deployment:
- can this person do these tasks with these tools and technologies
- can these tasks be performed under these conditions, to these standards and with this training?
It, therefore, links the technology with the task it is intended to support the people using it, their organisation and the environment. As such, a systems
perspective should be used from the outset of a project and should not be an afterthought. This will require involving various stakeholders including patients, carers and professionals in the technology deployment and development process. Technologies can be designed for stakeholders by informing, consulting or engaging them in the design process, but ideally, technologies can be codesigned with them, if possible. There can be many challenges in involving vulnerable patients or extremely busy healthcare practitioners, but designers have been creatively finding ways to achieve pragmatic and purposeful engagement of various stakeholders.
The panel discussion proposed that the effective deployment of digital pathology and integration of AI into the diagnostic workflow should be facilitated by working with colleagues with design expertise. This transdisciplinary approach recognises that the complex nature of technology deployment in healthcare is fundamentally people- and experience-centric. The subsequent integration of AI into the diagnostic workflow will also need to refocus attention – drawing it away from a purely technological focus to a more holistic systems approach that recognises how and even if AI is the appropriate technology for all diagnostic pathways.
In the Q&A session with the audience, we discussed the importance of being able to visualise AI tools so that pathologists and other users can better understand their operational status and validity. We also discussed how recreating a pathology environment and workflow that is more engaging would have benefits for staff and patients. A particular point of discussion was around the time people needed to innovate in their workplace. A contributor commented: “We can’t articulate what we want because we don’t have time to come out of the day job”.
It was interesting to hear input from Professor Cain and Dr Jun, who indicated that they must find creative design strategies to engage with busy people. This can be accomplished by creating engagement with smaller teams and planning at the outset to only take a short amount of people’s time during consultations. Professor Cain discussed her experience of engaging with time-poor staff; one of the approaches is that designers take their research to the time-poor staff and plan for studies to require short contributions from busy staff.
Professor Cain challenged us to ask the following question: ‘What could digital pathology achieve if it worked with others from different disciplines in a human-centred and co-creative way?’
The panel discussion was a conversation starter, and we hope that we can follow this up with more discussion among colleagues. We are happy to receive comments and to discuss joint projects in this area. It was certainly agreed that digital pathology and AI are only tools, and their impact will depend on how effectively we deploy them in clinical practice. We believe we can best do this collaboratively and with colleagues outside of our immediate disciplines.
A recording of the panel discussion is available here.
References are available in the original article.
Samar Betmouni, Consultant Neuropathologist, Sheffield Teaching
Hospitals NHS FT
Luisa Motta, Consultant Dermatopathologist Salford Royal NHS Trust
Gyuchan Jun, Reader in Socio-Technical System Design
School of Design & Creative Arts, Loughborough University
Rebecca Cain, Professor of Transdisciplinary Design
School of Design & Creative Arts, Loughborough University
Leave a Reply