Context
Cervical screening aims to prevent cervical cancer in women and people with a cervix by identifying and treating cell changes before they develop into cancer. Within the NHS Cervical Screening Programmes in the UK, samples are first tested for high-risk human papillomavirus (HPV). Samples that test positive are then assessed using cytology to identify cellular abnormalities that may require further investigation.
Delivering this service depends on a skilled laboratory workforce. However, cytology services across England, Scotland and Wales face workforce pressures, including high vacancy rates and an ageing workforce. These pressures may contribute to reporting delays and reduce the resilience of cervical screening services.
Artificial intelligence (AI)-assisted digital cytology has the potential to support laboratory staff. The Genius™ Digital Diagnostics System digitises cervical cytology slides and uses AI to identify and present areas of interest for review. The technology supports, rather than replaces, laboratory professionals, as these trained experts remain responsible for interpreting and reporting the results.
What we did
Aquarius Population Health researchers worked with laboratory experts in the UK to develop a decision-tree model representing the flow of cervical cytology slides through the NHS laboratory pathway.
They did an analysis that compared two approaches for cytology:
- Manual microscopy, representing the current standard approach
- AI-assisted digital cytology using the Genius™ system
The model followed the NHS Cervical Screening Programme laboratory workflow in England for one year. It simulated 479,125 HPV-positive samples requiring cytology triage, based on national screening volumes and HPV positivity rates.
The model included the time required for primary screening, rapid review, further checks and consultant reporting. It also accounted for the movement of physical slides within laboratories when manual microscopy was used. Published evidence, laboratory data and expert input informed the model parameters. The researchers also tested how uncertainty in key assumptions affected the results and explored an alternative laboratory workflow used in Scotland.
The analysis focused on staff time required to review and report slides. It did not assess the costs or wider practical requirements of implementing digital cytology, such as slide scanners, information technology infrastructure, training, validation and quality management.
Key findings
The model estimated that reviewing and reporting 479,125 cervical cytology slides annually would require:
- 31,842 staff hours using AI-assisted digital cytology
- 103,151 staff hours using manual microscopy.
This represented an estimated reduction of 71,309 staff hours.
The average time required to review and report each slide was estimated at:
- 4.0 minutes using AI-assisted digital cytology
- 12.9 minutes using manual microscopy.
Overall, the model indicated a potential 69% improvement in staff productivity. The largest estimated efficiency gain occurred during the primary screening step, where productivity increased by 76%. Consultant review productivity increased by an estimated 64%.
The alternative Scottish workflow also produced substantial estimated time savings, although the reduction was smaller than in the English base-case analysis. The uncertainty analysis showed that the time taken to screen and report slides was the main factor influencing the findings.
Implications
The findings suggest that the Genius™ Digital Diagnostics System could reduce the amount of laboratory staff time required to review and report cervical cytology slides. This may help services respond to workforce shortages, manage screening demand and support timely reporting of results.
However, the size of the benefit is likely to vary between laboratories according to local workflows, staffing arrangements and how the technology is implemented. A small proportion of slides may also remain unsuitable for digital assessment and require manual review.
The model did not examine diagnostic performance, downstream referrals or the wider costs and operational requirements of implementation. Prospective real-world studies in NHS laboratories are therefore needed to confirm the estimated time savings and assess broader effects on service capacity, backlogs and turnaround times.
To learn more about Aquarius Population Health’s work in health economic modelling, screening programmes and evidence generation, visit our website or contact us at info@aquariusph.com.