Three data-driven efficiency improvements for histopathology labs
by Matt Oates
Managers at NHS histopathology labs face increasing pressures to process diagnostics requests efficiently. Target priorities change constantly as new samples arrive with different service-level-agreement (SLA) turnaround times. In addition, targets for turnaround times are necessarily aggressive and the volume of test requests has increased. Consolidation of labs and transition to ‘hub and spoke’ organisational models make predicting the volume and variety of samples extremely challenging.
During my time as a research scientist supporting rare disease diagnosis, I’ve seen the good, the bad and the ugly impact that data science can have on lab efficiency. In this article, I’m sharing three ways in which you can significantly improve lab efficiencies through data science and software implementations that use existing lab systems and data assets.
As a starting point, the figure below shows a typical workflow for a histopathology lab and highlights some of the bottlenecks that can be addressed.
1. Integrate data assets
Sample data comes from multiple sources, and manual work is often required to match data from different systems to a sample. For example, your Laboratory Information Management System (LIMS) may use a different barcode and ID system to your sample tracking solution, while referral forms from GPs provide yet another ID for samples. Reception tasks such as entering sample data in the LIMS are sequential and labour-intensive. Using a software solution to integrate these types of data assets is a quick efficiency win.
Other information used in reception tasks might not be immediately recognised as data that you can use in a software solution. For example, clinical details in scanned handwritten notes or Word documents are often submitted as email attachments. Staff must read these documents individually before a sample can be adequately prioritised and scheduled. In fact, automated alignment of emails with LIMS records streamlines the process and accelerate information gathering.
If you have staff who are knowledgeable about software technology, you may be able to implement some of these changes yourselves. Usually, however, this kind of change is best implemented as part of an overall software solution that uses bespoke artificial intelligence (AI) algorithms tuned to address all your lab’s efficiency issues, with business rules that you can refine and change. With extensive experience in data analysis and software implementations, Unai is ideally placed to transform your lab’s efficiency.
2. Get more informed, predictive scheduling
Prioritising and scheduling sample processing depends on many factors. For example, cancer samples that require rapid assessment by a multi-disciplinary team (MDT) may have an SLA turnaround of a few days. In contrast, turnaround time for samples for a rare disease diagnosis could be a few weeks. Prioritisation needs to be reviewed on a daily basis as new requests come in and the workload changes.
One of the common barriers to efficiency is that manual and automated systems are used independently. Given the complexity and volume of requests to be processed, it is not realistic to expect managers to manually resolve detailed scheduling. On the other hand, fully-automated systems miss the nuances and opportunities that a highly skilled lab manager with years of experience can identify.
Processing variations, which depend on tissue type and sample size, can also impact workflow efficiency. Some samples take two hours of Tissue Processor machine time, whereas others require 12 hours or longer. Given these differences, managers may elect to simplify the daytime schedule by running some jobs overnight. This approach can lead to underutilisation of machines.
The good news is that AI-driven optimisation technology to assist managers in making informed scheduling decisions has progressed in recent years. New best-in-class tooling, such as Google OR-Tools, offers the intelligence and processing efficiency to help lab managers make more informed decisions on a day-to-day basis.
With our bespoke software solutions, Unai defines new open standards and data models in addition to using open-source core technologies. Consequently, labs don’t lock into a specific software tools’ vendor and retain control of their management information.
3. Factor in human nuances
Lab managers must allocate their specialist tissue processing teams and individual staff based on factors such as availability, skills and staff development goals. With the constant arrival of new samples, people planning is required on at least a daily basis.
In the early stages of a software solution deployment, it may be appropriate to focus on headline deliverables such as the 28-day cancer diagnosis target. As efficiencies are progressively embedded, managers can expand the reach of the solution to include detailed rules concerning scheduling constraints. For example:
1. 'Ensure all team members process at least 75% of MDT prioritised samples within 3 days of receipt.'
2. 'There is a glut of samples today for a specific organ system. Assign the most experienced person with those tissues from one of the other organ teams to help with the load.'
3. 'Bob is going on holiday. Make sure the member of staff with most similar experience profile covers his work.'
4. 'It's automatically noted Bob is the least assured on the team for a specific slide preparation from sample tracking. Both assign him slightly more of this work to practice in the next quarter and identify a mentorship pairing where Jane is efficient at this prep, but could learn from Bob.'
Unai’s software solutions incorporate business rules that put the manager in control of this type of operation. The software is there to support and inform a manager’s decisions. Consequently, managers can set the pace at which they move forward with efficiency improvements.
Realise the benefits
Data Science and software solutions can transform the work process in a histopathology lab. We estimate that software tuning of existing technology in NHS pathology labs can provide efficiency gains of at least 25% of current throughput levels.
Investing in the integration of your lab’s data assets not only saves time but also facilitates the extraction of maximum value from historic data. Data entry becomes the foundation of the lab’s goals instead of a tedious chore and barrier to more meaningful analysis of sample data.
Software solutions need to enable managers to learn how to best tune their lab's processes as they go. By focussing on resolving the key bottlenecks in current workflows, lab managers can progressively optimise their existing systems before significant additional investment in technology such as Digital Pathology.
Find out more
So, in an already over-tasked environment what can lab managers do to take the right first steps towards doing more with their existing data to realise the efficiencies they seek? To help answer our clients’ questions, Unai has codified 10 years of data science and data strategy knowledge into a diagnostic process that we call Discover.
Our Discover process follows a proven Ideation > Deep Dive > Solve approach in order to understand the client’s goals, resources and expected benefits. Discover is designed to be a rapid exercise that yields an immediate set of actionable “next steps” that is presented to the client in an accessible plain English format.
To start the process and arrange an informal initial discussion to better understand how Unai may be able to …