Is Pathology ready for the Digital and AI revolution?
The internet, Apps and AI have affected many aspects of our lives. Shopping, entertainment, communicating with social groups, booking travel, and banking are dramatically different experienced for my teenage children than they were for me. In spite of the many conferences, blogs, and £Billions of private equity funding, ‘Digital’ has had by comparison relatively limited effect to date on how we as citizens obtain healthcare, and how doctors and clinicians deliver healthcare.
There are many reasons why healthcare may be a slow adopter including clinical governance, legacy systems, and vested interests. Babylon’s recent launch of their GP App, and NHS England’s announcement of £45M to digitise GP surgeries, have dramatically increased the public awareness of digital health, and we may be at a tipping point.
Over the past eight years, I have spent significant time working in pathology. This is a heterogenous discipline which includes highly automated blood science laboratories processing up to 20,000 samples per day, many low volume sub specialist tests, and some areas which remain highly manual, such as histology, where samples are manually cut, processed into paraffin blocks, sliced thinly onto glass slides (sometimes with stains to detect cellular abnormalities), and interpreted under a microscope by a consultant, who may then attend MDT meetings in person to discuss the results.
The NHS has 200 consultant vacancies out of c2200 posts, and new training posts have not been filled for several years. District General Hospitals in remote areas in particular have struggled to fill posts. Histology reporting is often on the critical path to achieving cancer access times, and the NHS is spending c£60M per annum on backlog services and additional payments to consultants.
Whilst the process has made developments over the years, most 2017 histology labs would be recognisable to colleagues who retired 30 years ago. This may be about to change. A number of companies have been developing digital histology solutions for several years, but until recently the business cases didn’t stack up: scanners were slow and expensive, a single slide can be a 500MB file even with compression (c 100 MRI scans), and digital reading was slower than a microscope. Within the past twelve months, technology seems to have broken through, although many histopathology consultants remain reluctant to switch from microscopes.
I was recently invited to visit one of the early large scale adopters of digital histology. Not in the UK – I flew three hours to Bucharest to visit Synevo, the diagnostics division of Medicover. Many of my generation in the UK still associate the old Soviet nations with Trabants and crumbling concrete communist blocks. Romania has invested massively in technology infrastructure (my hotel broadband was 4x faster than any London hotel) and education.
The Synevo lab was a futuristic building, with well designed process flows, and better than most I have seen in Western Europe. Dr Stoicea, the lead histologist spent two hours talking passionately about quality, digital histology their investment in the Philips system, the journey to implement it, and his plans for future development.
Whilst some histologists see digital technology as a threat, it clearly could be used to improve outcomes and turnaround times:
Digital slides can be sent anywhere. Histologists can sub-specialise, which should result in greater accuracy of reporting.
Slides can be sent to where capacity is waiting. Why employ histopathology consultants? An Uber type system would allow the next available sub-specialist anywhere in the world to report.
Many slides are negative, but still need reporting. Many positive slides are stained, and the proportion of slides taking the stain are manually counted. Philips and others are developing AI systems which will look at slides and cells and undertake interpretation.
Some manual checking will probably be mandated for several years after these systems are developed, but they should reduce time per slide by guiding consultants to anomalous cells, and doing routine counting. BD’s focal point technology already offers this in gynaecological cytology.
Quality control can be improved.The system can track what the consultant looked at, potentially prompt them if they missed something, and allow them to annotate and send to other global experts for a second opinion.
Slides can be reported anywhere.Staff who would otherwise take career breaks could report from home.
MDT meetings can be virtual, again with a global histology expert attending for the part of the meeting where he is needed.
There are very substantial discrepancies between the salaries of histologists around the world – probably ranging from £30K to >£300K within Europe. The efficiency of histologists in terms of comparable cases reported also varies by a factor of 3-4 between the least efficient public sector services and the most efficient private ones. The correlation between efficiency and salary is also far from perfect.
It is difficult to see how this economic disparity can be sustained, once slides can move around the globe at a click. If China were to train 10,000 English speaking histology consultants, they could dominate then world in reporting, as they have done in manufacturing, and as they seem intent upon doing in genetic testing, based upon their investment in genetic sequencing labs.
CEO, Latchmore Associates
Founded in 2003, Latchmore Associates supports public, private and third sector health and social care organisations, with particular expertise in the following areas:
• Supporting overseas/ new entrants to the UK market • NHS private patients, commercialisation and revenue generation • Public/private partnerships from market sounding through to procurement and implementation • Pathology partnerships/ consolidation • Bid and transaction negotiation support • Interim management