JMIR Publications

ABSTRACT

Background:

Despite widespread concern regarding new technology replacing jobs, or how technology will change the structure of jobs, we lack detailed real-world evidence about what can and cannot be automated. This lack of understanding can be confusing and dangerous for policymakers who often see healthcare-oriented roles as “low risk” for automation. However, within UK primary healthcare, automation is not regarded as a threat to staff, but as a necessary solution to tackle many issues it currently faces, such as staff shortages, increased demand and reduced budget to name a few.

Objective:

The purpose of this research is to understand the current state of automation, and the potential opportunities and challenges of further automation in the UK primary healthcare sector. Our first aim is to observe and collate a comprehensive understanding of what occupations and tasks occur in primary healthcare practices. No such similar dataset currently exists for analysing the work practices and tasks of NHS primary care staff. Our second aim is to use expert knowledge of the current state-of-the-art automation technologies as a guide to estimate what tasks and work practices could be susceptible to further automation opportunities going forward, and the potential effects on healthcare workflows.

Methods:

This project utilises a multi-method and mixed-method research design, comprising of two phases: a qualitative observational phase, and a quantitative data analysis phase; each phase addressing one of the two project aims. A critical part of the problem we propose to address is that a formal framework for measuring automation is somewhat lacking in the literature. The healthcare domain offers a further challenge in measuring automation because of a general lack of detailed, healthcare-specific, occupation and task observational data to form good insights about this notoriously miss-understood topic. Therefore, our first aim focuses on addressing this lack of data by collecting high quality, detailed task-specific data from UK primary healthcare practices. The second aim then proposes a formal framework for probabilistic inference of task and occupation-automation to gain valuable insights.

Results:

Our detailed fieldwork includes observing and documenting 16 unique occupations performing over 160 tasks across 7 primary care centres. Our initial findings are that tasks are often shared amongst staff and can include convoluted workflows which often vary between practices. The single most-used technology in primary healthcare is the desktop computer. Secondly, we have conducted a large-scale survey of 156 machine learning and robotics experts to assess what tasks are susceptible to automation given the current state-of-the-art technology available today. Further results and detailed analysis will be published towards the end of the project.

Conclusions:

To date, tasks observed in primary healthcare can be categorised into two types: structured or unstructured. We believe our analysis, will identify many of the structured tasks to be highly susceptible to automation using current technology, given that sufficient data can be collected. Further, we aim to identify workflows that, given the proper implementation of automation technologies, could save considerable staff resources.