When Logan Tod & Co started this journey to support the NHS reduce elective waiting during the start of the Covid-19 pandemic, we wanted to do things differently with data and analytics.
For many years, we noticed NHS business users including executive, management and delivery teams finding it difficult to extract value from their data. The analytical tools used can be complex for non-analysts and the analysis generated provides oversight rather than relevant actionable insights.
We set ourselves a challenge to find new techniques and tools that analysts could easily adopt to make immediate impact to business users. One such discovery is AI-enhanced process mining as a proven technique to analyse any process flow, with widespread adoption outside the NHS.
What is process mining?
Process mining is a technique to analyse and monitor processes. A patient journey can be viewed as a series of process steps e.g. in elective care, patients start their journey from a referral, flowing through a series of processes before completing their journey with a discharge.
Currently in most NHS organisations, process workshops and interviews are conducted to understand processes, which results in an idealised picture of a process. This approach takes time and often require significant inputs from a large team of experts.
Using AI-enhanced process mining tools, and data from any patient tracking list, teams can speed up the analytics process using existing data and reveal what is really happening on the ground.
We have included below a snapshot of a proof of concept with the AnalystX community. Using national referral to treatment (RTT) data dictionary to define the process steps for open pathways, we can see patients with a process start at RTT Start and their different paths to where their journey has come to a stop (process end).
What can I do with the AI-enhanced process mining tools?
- Understand process breakdowns - where patient flows are disrupted
- Find 'lost' patients that are stuck in limbo, or having processes duplicated
- Spot the real bottlenecks and know what to invest in - clinical, admin or diagnostic capacity
- Remove administrative burdens
- Release capacity by using the data to re-imagine patient flow
- Demonstrate the real patient experience and understand the impact on outcomes
- Spot inequality - see which patient groups wait longer
- Measure the impact of non-conformance to best practice
- Drive a data driven continuous improvement process
The most important output is being able to visualise the real patient journey so that
All teams involved can easily understand what they are looking at, get value from their own data, and start identifying appropriate actions to accelerate patient flow.
Once everybody can see the reality of the patient journey it is much easier to collaborate and deliver real improvement. In this example the impact of a cancelled appointment is to add 39 days on average to the patient journey, but there are also some variants in the process to understand, for example multiple RTT starts.
Can it work in the NHS?
We have now applied AI-enhanced process mining to analyse patient flow in discharge to assess (d2a) and elective care. We have demonstrated that with any patient flow data, analysts will be able to provide actionable operational insights to business users in days, sometimes within hours, not weeks or months that many NHS business users would expect.
How can the NHS adopt this technique?
One of the attractions of AI-enhanced process mining is that it does not require significant data science skills, and is truly accessible to all. The key element of any successful analysis is data preparation - in process mining, this relates to the preparation of Event logs where a single entry is an unique event ID, an associated activity and the timestamp of the activity. All other information about an event are considered optional attributes that can be easily linked to the Event log using the unique event ID to create an incredibly rich and relevant analysis.
Examples of optional attributes include:
Patient attribute - age, gender, disability
Workforce attributes - specialism, staff group, working hours
Outcome attributes - referral accepted, discharged, follow-up appointment required
Pathway attributes - RTT, Cancer 2-weeks wait, Urgent, Routine, Admitted, Non-Admitted
Both the Event log and optional attributes data typically resides in existing systems within the NHS e.g. PTL, PAS, EPR which means AI-enhanced process mining can be applied to any NHS organisation today.
How long does it take and what does it cost?
If you are using a specialised process mining tool, and you have a prepared data set then the answer is days to get actionable insights. Analysts will also be able to engage business users in data interrogation and process improvements immediately.
Specialised process mining tools include Celonis and Apromore. Other generic tools to which process mining functionality can be added, at a cost, include PowerBI and Tableau. For those with good programming skills, R Studio and Python also offer means to conduct process mining.
We tried a good range of both specialised tools and generic tools. Our clear preference is the use of specialised process mining tool like Celonis. The features and functions are purpose built but yet provides a wide range of additional capabilities, including animation, push notifications, scenario planning etc. Most importantly, it really engages business users visually, and the NHS can get started using a free version like we did. Once it is proven, you can then decide on the most appropriate tool and paid-for functionality required.
So to answer the question directly, you can demonstrate value within two weeks at limited or no cost, subject to getting appropriate anonymised Event log and relevant attributes.
What else do you need to consider?
AI-enhanced process mining has the ability to create incredibly rich patient-centric insights that will power improvements to patient journeys including waiting time, quality of care and equality of access. To achieve such powerful insights, a wide spectrum of data about a patient journey has to be linked for analysis. Therefore, do consider how you can anonymise data for insights but still enable business users to take specific actions for each individual patient using the powerful insights.
We would advise initial investigations using fully anonymised data, following the relevant data processing standards or guidelines that apply to your organisation. Our experience has proven that anonymised data can generate a wealth of insight and engagement, enabling teams to make meaningful change.
How to get started?
There are a few options:
- Search for AI-enhanced process mining in your web browser and see what comes up, there are many examples from organisations outside the NHS you could learn from
- Get in touch with us here, we can share relevant NHS case studies
- Join AnalystX where we will post more information and host demonstrations over the coming months
- Register interest with us to be part of the next wave of pilot projects as we seek to expand and extend the use of process mining
What does Logan Tod & Co offer?
We are a not-for-profit organisation seeking to improve outcomes for ordinary citizens and patients. We do this by supporting the development of new analytical techniques in organisations delivering social value like the NHS. We can support you in a number of ways:
- Analyse your data for you - a great way to get started with process mining and understand the impact it could make in your organisation
- Coach your team to do the analysis - our long term goal is to help develop capability so we are only too happy to take on a supporting role to help build skills and knowledge within your team
- Build a tailored solution for you if you do not have time and resources to develop and implement a new solution
- Deliver a complete outsourced solution if you just want to focus on patient acceleration and getting through the backlog
Just get in touch to talk about AI-enhanced process mining or give us feedback - we love meeting people!